DeepSeek: DeepSeek V4.1 Flash
DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on …
deepseek/deepseek-v4.1-flash-20260910
DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company's Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on output from a 552B-parameter backbone, an asymmetric split that keeps per-token compute low relative to the model's total size. Image understanding is native to the architecture, with visual and text embeddings trained jointly from the start of pre-training rather than added afterward as in the earlier experimental [V4 Flash Vision Exp](https://openrouter.ai/deepseek/deepseek-v4-flash-vision-exp).
It is suited for coding, terminal, and computer-use agents, along with long-horizon tasks that must run to completion across many steps and long-context analysis. Compressed KV caching cuts cache memory to roughly a quarter of the previous Flash generation, significantly reducing costs on agentic workloads. DeepSeek positions it as the cost-efficient tier of the V4.1 family and reports that it exceeds [V4 Pro](https://openrouter.ai/deepseek/deepseek-v4-pro-0813) on performance, speed, and task completion time.
Specifications & capabilities
- Input
- text, image
- Output
- text
- Context window · tokens
- 1048576
- Published model ID
- DeepSeek-V4.1-Flash
- Capabilities
- multimodal, promptCaching, reasoning, structuredOutputs, toolCalling, visionInput
Description source ↗Evaluations
41–60 / 83
72.889%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · ModelRun
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 6
- sample size
- 150
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 3
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- ModelRun
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-28T06:40:49.113Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 150
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7288890000000001
- accuracy_std_dev
- 0.03142665378305492
- cost_per_task_usd
- 0.02386675333333333
- output_tokens_per_task
- 22554.02666666667
- latency_seconds_per_task
- 151.22852
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 8
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗71.178%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · OpenInference
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 6
- sample size
- 447
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 9
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- OpenInference
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T13:23:14.602Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 447
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7117822371364653
- accuracy_std_dev
- 0.056118924341169225
- cost_per_task_usd
- 0.0462677941834452
- output_tokens_per_task
- 60076.49888143177
- latency_seconds_per_task
- 1033.6222237136465
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 6
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗71.111%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Cohere
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 6
- sample size
- 150
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 3
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Cohere
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:01:01.015Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 150
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7111113333333333
- accuracy_std_dev
- 0.006285236475706792
- cost_per_task_usd
- 0.06754876666666666
- output_tokens_per_task
- 33581.94
- latency_seconds_per_task
- 139.69535333333334
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 7
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗90.44%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · DeepInfra
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 7
- sample size
- 1977
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 10
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- DeepInfra
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T08:03:09.643Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 1977
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.9044004522003033
- accuracy_std_dev
- 0.00856689234728672
- cost_per_task_usd
- 0.014848743045017704
- output_tokens_per_task
- 28855.582195245323
- latency_seconds_per_task
- 370.3765558927668
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 71
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗89.289%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Wafer
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 7
- sample size
- 3316
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 17
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Wafer
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T08:57:07.688Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 3316
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.8928929372738239
- accuracy_std_dev
- 0.015585617718297187
- cost_per_task_usd
- 0.014895325693606755
- output_tokens_per_task
- 21447.96381182147
- latency_seconds_per_task
- 187.17530186972255
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 13
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗87.984%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Venice
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 7
- sample size
- 1584
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 8
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Venice
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T09:14:35.460Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 1584
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.87984
- accuracy_std_dev
- 0.018052164925570553
- cost_per_task_usd
- 0.036141941919191914
- output_tokens_per_task
- 28081.721590909092
- latency_seconds_per_task
- 139.65641224747475
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 39
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗87.673%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Relace
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 7
- sample size
- 2969
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 15
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Relace
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-11-02T17:21:00.007Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 2969
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.8767260400808354
- accuracy_std_dev
- 0.008082492995494786
- cost_per_task_usd
- 0.024447516335466487
- output_tokens_per_task
- 34286.534523408554
- latency_seconds_per_task
- 220.1124371842371
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 13
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗87.433%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · BaseTen
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 7
- sample size
- 4074
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 21
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- BaseTen
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T09:03:20.938Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 4074
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.8743250351006383
- accuracy_std_dev
- 0.026296453097608783
- cost_per_task_usd
- 0.03975532817869414
- output_tokens_per_task
- 25880.549828178693
- latency_seconds_per_task
- 80.92118360333824
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 16
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73.923%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · InferenceNet
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 7
- sample size
- 650
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 13
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- InferenceNet
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-11-04T21:17:38.304Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 650
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7392308461538462
- accuracy_std_dev
- 0.0224239080065011
- cost_per_task_usd
- 0.017090104615384617
- output_tokens_per_task
- 22554.26307692308
- latency_seconds_per_task
- 299.6744630769231
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 10
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73.8%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Morph
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 7
- sample size
- 500
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 10
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Morph
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:03:50.388Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 500
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7380000999999999
- accuracy_std_dev
- 0.02844855708274146
- cost_per_task_usd
- 0.016056571999999998
- output_tokens_per_task
- 17634.696
- latency_seconds_per_task
- 534.586188
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 9
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73.593%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Makora
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 7
- sample size
- 450
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 9
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Makora
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:16:10.183Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 450
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7359261111111111
- accuracy_std_dev
- 0.02318882707227026
- cost_per_task_usd
- 0.04132428666666666
- output_tokens_per_task
- 25779.64
- latency_seconds_per_task
- 135.74828
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 11
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73.292%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Modal
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 7
- sample size
- 400
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 8
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Modal
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:25:59.335Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 400
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.732916625
- accuracy_std_dev
- 0.018888392049996622
- cost_per_task_usd
- 0.0482453475
- output_tokens_per_task
- 29218.5525
- latency_seconds_per_task
- 162.563185
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 11
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗86.973%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Fireworks
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 8
- sample size
- 3347
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 17
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Fireworks
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T08:35:39.353Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 3347
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.8697338870630417
- accuracy_std_dev
- 0.011260986943785035
- cost_per_task_usd
- 0.04833613654018524
- output_tokens_per_task
- 37254.27158649537
- latency_seconds_per_task
- 200.5837881685091
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 17
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗75%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · SiliconFlow
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 8
- sample size
- 300
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 6
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- SiliconFlow
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:28:07.595Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 300
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7499999999999999
- accuracy_std_dev
- 0.03349968490896593
- cost_per_task_usd
- 0.028200379999999997
- output_tokens_per_task
- 15323.823333333334
- latency_seconds_per_task
- 163.95521
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 28
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗74.519%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · StreamLake
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 8
- sample size
- 450
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 9
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- StreamLake
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:25:02.563Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 450
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7451853333333333
- accuracy_std_dev
- 0.031234315132203188
- cost_per_task_usd
- 0.01557802666666667
- output_tokens_per_task
- 15042.41111111111
- latency_seconds_per_task
- 147.55571333333336
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 25
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73.494%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Fireworks
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 8
- sample size
- 747
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 15
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Fireworks
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:41:59.422Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 747
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7349398099062917
- accuracy_std_dev
- 0.02455646627618342
- cost_per_task_usd
- 0.05549164792503348
- output_tokens_per_task
- 30263.555555555555
- latency_seconds_per_task
- 157.9741485943775
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 14
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗73%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · Nebius
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 8
- sample size
- 250
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 5
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- Nebius
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T11:52:51.154Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 250
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.73
- accuracy_std_dev
- 0.0268328157299975
- cost_per_task_usd
- 0.074570136
- output_tokens_per_task
- 30037.412
- latency_seconds_per_task
- 246.781984
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 23
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗72.704%
- source id
- openRouterBenchmarks
- leaderboard name
- tau2-bench Verified Airline · Accuracy · BaseTen
- dimension
- agentic
- metric
- tau2-bench Verified Airline · Accuracy
- rank
- 8
- sample size
- 392
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 8
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- tau_bench_verified_airline
- name
- tau2-bench Verified Airline
- record_kind
- aggregate
- provider_name
- BaseTen
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T12:26:53.018Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- tau2-bench Verified Airline · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 392
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/tau2-bench-airline
- measurements
- accuracy
- 0.7270409540816326
- accuracy_std_dev
- 0.02754153279934688
- cost_per_task_usd
- 0.04606436989795917
- output_tokens_per_task
- 19891.04081632653
- latency_seconds_per_task
- 81.57688520408163
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 14
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗88.653%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Nebius
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 9
- sample size
- 990
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 5
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Nebius
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T08:28:17.984Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 990
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.8865318000000001
- accuracy_std_dev
- 0.007110736653821441
- cost_per_task_usd
- 0.030907244444444443
- output_tokens_per_task
- 25534.58484848485
- latency_seconds_per_task
- 199.10552626262626
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 25
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗88.237%
- source id
- openRouterBenchmarks
- leaderboard name
- GPQA Diamond · Accuracy · Together
- dimension
- knowledge
- metric
- GPQA Diamond · Accuracy
- rank
- 9
- sample size
- 1584
- aggregation scope
- status
- published
- harness
- url
- https://github.com/OpenRouterTeam/benchmark-harness
- name
- OpenRouter benchmark harness
- direction
- higherIsBetter
- run_count
- 8
- conditions
- window
- newest 90 days; successful visible nonzero runs
- dataset
- benchmark-leaderboard
- benchmark
- id
- gpqa_diamond
- name
- GPQA Diamond
- record_kind
- aggregate
- provider_name
- Together
- attribution
- Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
- last_run_at
- 2026-10-31T09:02:59.407Z
- license_url
- https://creativecommons.org/licenses/by/4.0/
- methodology
- Rows aggregate the benchmark-runs dataset exactly as the leaderboard pages do, over each benchmark's non-zero-score runs from the newest 90 days: accuracy is the task-weighted mean of per-run accuracy, accuracy_std_dev is the population standard deviation of per-run accuracy (null with a single run), and the per-task averages divide summed cost, generation time, and output tokens by summed tasks. Rows are ordered by accuracy descending, then model name.
- metric_kind
- GPQA Diamond · Accuracy
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- sample_size
- 1584
- source_file
- benchmark-leaderboard.json
- source_page
- https://openrouter.ai/benchmarks/gpqa-diamond
- measurements
- accuracy
- 0.882365125
- accuracy_std_dev
- 0.0043279540904883765
- cost_per_task_usd
- 0.029606650252525252
- output_tokens_per_task
- 24519.371843434343
- latency_seconds_per_task
- 78.74681123737373
- published_at
- 2026-11-16T01:46:02.653Z
- snapshot_date
- 2026-11-16
- source_revision
- 7ccaf17664252a3053dcb7df8ad0a46e72954a3b4524ddc1c8a7ef8a41b15568
- transformations
- TensorTrace preserves source results, converts latency to seconds and derives ranks within matching conditions. Individual runs have no derived rank.
- metric_population
- 25
- benchmark_evaluation
- true
- openrouter_permaslug
- deepseek/deepseek-v4.1-flash-20260910
- source model id
- deepseek/deepseek-v4.1-flash-20260910
Original result ↗Published prices
1–20 / 96
$0.0042
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- cache_read_tokens_per_million
- Currency
- USD
$0.14
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- input_tokens_per_million
- Currency
- USD
$0.42
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- output_tokens_per_million
- Currency
- USD
$0.0034
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- mode
- batch
$0.11
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- mode
- batch
$0.34
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- output_tokens_per_million
- Currency
- USD
- Billing conditions
- mode
- batch
$0.0030
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:off:14-24
- label
- 14:00 – 24:00(UTC)
- Billing label
- Cached input tokens (per 1M)
$0.01
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:on:0-14
- label
- 00:00 – 14:00(UTC)
- Billing label
- Cached input tokens (per 1M)
$0.15
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:off:14-24
- label
- 14:00 – 24:00(UTC)
- Billing label
- Input tokens (per 1M)
$0.30
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:on:0-14
- label
- 00:00 – 14:00(UTC)
- Billing label
- Input tokens (per 1M)
$0.60
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- output_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:off:14-24
- label
- 14:00 – 24:00(UTC)
- Billing label
- Output tokens (per 1M)
$1.20
- Inference provider
- siliconFlow
- Platform
- SiliconFlow
- Unit
- output_tokens_per_million
- Currency
- USD
- Billing conditions
- tier
- peak:on:0-14
- label
- 00:00 – 14:00(UTC)
- Billing label
- Output tokens (per 1M)
$0.02
- Inference provider
- vercelAIGatewayCatalog
- Platform
- alibaba
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- alibaba | deepseek/deepseek-v4.1-flash
- model_type
- language
- peak_pricing
- windows
- end_minute_utc
- 840
- start_minute_utc
- 0
- multiplier
- 2
- billing_scope
- via Vercel AI Gateway
- Billing label
- Cached input / 1M
- Pricing via
- Vercel AI Gateway
$0.15
- Inference provider
- vercelAIGatewayCatalog
- Platform
- alibaba
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- alibaba | deepseek/deepseek-v4.1-flash
- model_type
- language
- peak_pricing
- windows
- end_minute_utc
- 840
- start_minute_utc
- 0
- multiplier
- 2
- billing_scope
- via Vercel AI Gateway
- Billing label
- Input tokens / 1M
- Pricing via
- Vercel AI Gateway
$0.60
- Inference provider
- vercelAIGatewayCatalog
- Platform
- alibaba
- Unit
- output_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- alibaba | deepseek/deepseek-v4.1-flash
- model_type
- language
- peak_pricing
- windows
- end_minute_utc
- 840
- start_minute_utc
- 0
- multiplier
- 2
- billing_scope
- via Vercel AI Gateway
- Billing label
- Output tokens / 1M
- Pricing via
- Vercel AI Gateway
$0.01
- Inference provider
- vercelAIGatewayCatalog
- Platform
- baseten
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- baseten | deepseek/deepseek-v4.1-flash
- model_type
- language
- billing_scope
- via Vercel AI Gateway
- Billing label
- Cached input / 1M
- Pricing via
- Vercel AI Gateway
$0.01
- Inference provider
- vercelAIGatewayCatalog
- Platform
- baseten
- Unit
- cache_read_tokens_per_million
- Currency
- USD
- Billing conditions
- region
- scope
- zone
- geo_region
- us
- endpoint
- baseten | deepseek/deepseek-v4.1-flash
- model_type
- language
- billing_scope
- via Vercel AI Gateway
- Billing label
- Cached input / 1M
- Pricing via
- Vercel AI Gateway
$0.30
- Inference provider
- vercelAIGatewayCatalog
- Platform
- baseten
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- baseten | deepseek/deepseek-v4.1-flash
- model_type
- language
- billing_scope
- via Vercel AI Gateway
- Billing label
- Input tokens / 1M
- Pricing via
- Vercel AI Gateway
$0.30
- Inference provider
- vercelAIGatewayCatalog
- Platform
- baseten
- Unit
- input_tokens_per_million
- Currency
- USD
- Billing conditions
- region
- scope
- zone
- geo_region
- us
- endpoint
- baseten | deepseek/deepseek-v4.1-flash
- model_type
- language
- billing_scope
- via Vercel AI Gateway
- Billing label
- Input tokens / 1M
- Pricing via
- Vercel AI Gateway
$1.20
- Inference provider
- vercelAIGatewayCatalog
- Platform
- baseten
- Unit
- output_tokens_per_million
- Currency
- USD
- Billing conditions
- endpoint
- baseten | deepseek/deepseek-v4.1-flash
- model_type
- language
- billing_scope
- via Vercel AI Gateway
- Billing label
- Output tokens / 1M
- Pricing via
- Vercel AI Gateway
Provider specifications & access
1–6 / 6
- source model id
- azure-huggingface/deepseek-ai--deepseek-v4.1-flash@1
- hosting
- foundry-managed-compute
- license
- MIT
- context length
- 1048576
- source model id
- azure-huggingface/deepseek-ai--deepseek-v4.1-flash@2
- hosting
- foundry-managed-compute
- license
- MIT
- context length
- 1048576
- source model id
- deepseek-ai/deepseek-v4.1-flash
- hosting
- nvidia-catalog
- context length
- 1048576
- source model id
- deepseek-ai/DeepSeek-V4.1-Flash
- hosting
- siliconflow-hosted
- max output tokens
- 393216
- capabilities
- promptCaching
- reasoning
- structuredOutputs
- toolCalling
- source model id
- deepseek/deepseek-v4.1-flash
- hosting
- vercel-gateway
- context length
- 1048576
- max output tokens
- 32768
- capabilities
- promptCaching
- reasoning
- structuredOutputs
- toolCalling
- visionInput
- source model id
- publishers/deepseek-ai/models/deepseek-v4.1@deepseek-v4.1-flash
- hosting
- vertex-self-deployed
Safety assessments
1–1 / 1
32 / 100
- source id
- stanfordFMTI
- canonical model id
- deepseek/deepseek-v4.1-flash-20260910
- taxonomy
- transparency
- hazard category
- transparency
- hazard name
- Organization transparency
- policy area
- DeepSeek
- assessment type
- organization-level disclosure
- metric
- disclosure_score
- result label
- 32 / 100
- sample size
- 100
- updated at
- 2026-11-09T18:08:05.227Z
Source specification observations
1–11 / 11
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- azureFoundry
- confidence
- 1
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- azureFoundry
- confidence
- 1
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- nvidiaNim
- confidence
- 1
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- openRouter
- confidence
- 1
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- openRouter
- confidence
- 1
- field
- context_length
- value
- 1049K
- scope
- source-model
- source id
- siliconFlow
- confidence
- 1
- field
- context_length
- value
- 1048576
- normalized value
- 1048576
- scope
- source-model
- source id
- vercelAIGatewayCatalog
- confidence
- 1
- field
- max_output_tokens
- value
- 393216
- normalized value
- 393216
- scope
- source-model
- source id
- openRouter
- confidence
- 1
- field
- max_output_tokens
- value
- 131072
- normalized value
- 131072
- scope
- source-model
- source id
- openRouter
- confidence
- 1
- field
- max_output_tokens
- value
- 393K
- scope
- source-model
- source id
- siliconFlow
- confidence
- 1
- field
- max_output_tokens
- value
- 32768
- normalized value
- 32768
- scope
- source-model
- source id
- vercelAIGatewayCatalog
- confidence
- 1
Hugging Face model card
1–1 / 1
- downloads
- 1694181
- likes
- 4978
- pipeline tag
- image-text-to-text
- library name
- transformers
- License
- mit
Explore more models