MiniMax: MiniMax M3

MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding, a…

minimax/minimax-m3-20260531

MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding, and tool use. It is built on MiniMax Sparse Attention (MSA), which replaces full attention with KV-block selection to cut per-token compute at long context — roughly 1/20 the cost of the previous generation at 1M tokens, with substantially faster prefill and decode while retaining quality across most tasks.

Trained as a native multimodal model on interleaved data and tuned for multi-turn, production-like collaboration via an interactive user-simulator framework, the model is oriented toward sustained, multi-step tasks rather than single-turn execution.

Specifications & capabilities

Input
text, image, video
Output
text
Context window · tokens
1048576
Published model ID
MiniMax-M3
Capabilities
multimodal, promptCaching, reasoning, streaming, structuredOutputs, toolCalling, videoInput, visionInput
Description source ↗

Evaluations

1–20 / 52

OpenRouter Benchmarks

91.659%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Minimax
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
1
sample size
2178
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
11
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
Minimax
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:05:09.530Z
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
2178
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9165900909090909
accuracy_std_dev
0.009424357296191685
cost_per_task_usd
0.0340643388429752
output_tokens_per_task
28227.440771349862
latency_seconds_per_task
223.57000459136822
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
4
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.802%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Novita
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
1
sample size
2178
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
11
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
Novita
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:09:19.241Z
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
2178
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9080193636363636
accuracy_std_dev
0.0188735967823871
cost_per_task_usd
0.037270338383838385
output_tokens_per_task
31082.94582185491
latency_seconds_per_task
248.01031542699727
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
59
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.539%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · ModelRun
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
1
sample size
2970
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
ModelRun
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:10:09.223Z
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
2970
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9053869333333332
accuracy_std_dev
0.006878297322900644
cost_per_task_usd
0.07875497138047138
output_tokens_per_task
31017.457912457914
latency_seconds_per_task
227.26118787878787
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

83.461%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Mara
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
1
sample size
782
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
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
Mara
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:08:31.063Z
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
782
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.8346120933503837
accuracy_std_dev
0.029860364128538167
cost_per_task_usd
0.09530405242966751
output_tokens_per_task
39472.66240409207
latency_seconds_per_task
330.0654283887468
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
5
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

74.254%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · Mara
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
1
sample size
391
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
Mara
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T19:27:07.786Z
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
391
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/tau2-bench-airline
measurements
accuracy
0.7425405959079282
accuracy_std_dev
0.026283120985422872
cost_per_task_usd
0.12785004603580563
output_tokens_per_task
10880.268542199488
latency_seconds_per_task
203.56263938618926
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
2
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

73.467%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · Minimax
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
1
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
Minimax
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T19:09:49.337Z
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
0.7346668
accuracy_std_dev
0.03066679130525396
cost_per_task_usd
0.029810941999999993
output_tokens_per_task
10584.172
latency_seconds_per_task
101.760856
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
4
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

91.498%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Together
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
2
sample size
1980
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
Together
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:17:15.690Z
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
1980
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9149829999999999
accuracy_std_dev
0.006208509031965745
cost_per_task_usd
0.035431615151515154
output_tokens_per_task
29385.477272727272
latency_seconds_per_task
289.87857626262627
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.769%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · StreamLake
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
2
sample size
2376
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
12
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
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
2026-10-31T16:23:41.132Z
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
2376
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9076879166666667
accuracy_std_dev
0.026811471259078443
cost_per_task_usd
0.036969556397306404
output_tokens_per_task
30839.63425925926
latency_seconds_per_task
266.792257996633
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
30
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.733%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · CoreWeave
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
2
sample size
2374
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
12
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
CoreWeave
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-31T16:32:36.327Z
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
2374
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9073293664700928
accuracy_std_dev
0.008220232321331707
cost_per_task_usd
0.030960309604043806
output_tokens_per_task
32028.098146588036
latency_seconds_per_task
400.8752552653749
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
27
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.598%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · GMICloud
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
3
sample size
2574
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
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
GMICloud
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-11-10T06:30:46.502Z
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
2574
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9059826923076922
accuracy_std_dev
0.02531585573200691
cost_per_task_usd
0.03357483760683761
output_tokens_per_task
27670.721056721057
latency_seconds_per_task
222.9692094017094
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
26
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

74.5%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · SambaNova
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
3
sample size
200
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
4
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
SambaNova
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-17T19:47:21.934Z
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
200
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/tau2-bench-airline
measurements
accuracy
0.745
accuracy_std_dev
0.026822730826670128
cost_per_task_usd
0.15284815
output_tokens_per_task
28246.935
latency_seconds_per_task
494.52749
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
5
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

91.405%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · DeepInfra
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
4
sample size
892
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
7
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
2026-10-31T16:03:21.470Z
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
892
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9140508968609865
accuracy_std_dev
0.01852693073581796
cost_per_task_usd
0.02261122085201794
output_tokens_per_task
20985.829596412557
latency_seconds_per_task
825.5830426008969
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.618%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Venice
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
4
sample size
2169
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
11
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
2026-10-31T16:12:28.227Z
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
2169
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9061779082526508
accuracy_std_dev
0.011623858653648528
cost_per_task_usd
0.026847682342093126
output_tokens_per_task
22212.46196403873
latency_seconds_per_task
336.78801567542644
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

90.202%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Parasail
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
4
sample size
2145
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
12
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
Parasail
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-11-06T06:01:51.918Z
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
2145
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.9020202237762238
accuracy_std_dev
0.008897393508385223
cost_per_task_usd
0.029685459207459214
output_tokens_per_task
23066.490442890445
latency_seconds_per_task
182.81628065268066
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
36
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

81.921%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · SambaNova
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
4
sample size
909
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
gpqa_diamond
name
GPQA Diamond
record_kind
aggregate
provider_name
SambaNova
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-17T16:06:46.637Z
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
909
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.8192149207920791
accuracy_std_dev
0.018823182524040232
cost_per_task_usd
0.11224348184818483
output_tokens_per_task
46532.80198019802
latency_seconds_per_task
401.07960946094613
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

76.583%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · Together
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
4
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
Together
attribution
Source: OpenRouter (openrouter.ai), licensed under CC BY 4.0: reuse and republish with attribution to OpenRouter.
last_run_at
2026-10-24T19:16:17.608Z
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
0.76583325
accuracy_std_dev
0.03273434952672042
cost_per_task_usd
0.0269725475
output_tokens_per_task
9025.64
latency_seconds_per_task
160.221945
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
22
benchmark_evaluation
true
openrouter_permaslug
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

74.19%

source id
openRouterBenchmarks
leaderboard name
tau2-bench Verified Airline · Accuracy · ModelRun
dimension
agentic
metric
tau2-bench Verified Airline · Accuracy
rank
5
sample size
700
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
14
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
2026-10-31T19:13:33.926Z
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
700
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/tau2-bench-airline
measurements
accuracy
0.7419049285714285
accuracy_std_dev
0.027305278452197487
cost_per_task_usd
0.05683837428571428
output_tokens_per_task
10137.55
latency_seconds_per_task
141.2787842857143
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenRouter Benchmarks

86.532%

source id
openRouterBenchmarks
leaderboard name
GPQA Diamond · Accuracy · Morph
dimension
knowledge
metric
GPQA Diamond · Accuracy
rank
6
sample size
198
aggregation scope
status
published
harness
url
https://github.com/OpenRouterTeam/benchmark-harness
name
OpenRouter benchmark harness
direction
higherIsBetter
run_count
1
conditions
window
newest 90 days; successful visible nonzero runs
dataset
benchmark-leaderboard
benchmark
id
gpqa_diamond
name
GPQA Diamond
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
2026-08-22T10:03:34.994Z
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
198
source_file
benchmark-leaderboard.json
source_page
https://openrouter.ai/benchmarks/gpqa-diamond
measurements
accuracy
0.86532
cost_per_task_usd
0.02509249494949495
output_tokens_per_task
25370.131313131315
latency_seconds_per_task
884.8438888888888
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
minimax/minimax-m3-20260531
source model id
minimax/minimax-m3-20260531
Original result ↗

OpenCompass LLM

90.8%

source id
openCompass
leaderboard name
Knowledge_Science
dimension
general
metric
Knowledge_Science
rank
8
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
KnowledgeTable
catalog
closed
snapshot
month
26-07
fileName
assets/llm-rank/llm-data-v2.26-07.20260824.json
updateTime
2026-08-24
columnFileName
assets/llm-rank/llm-column-v2.26-07.20260824.json
metric_kind
Knowledge_Science
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
1-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
15
benchmark_evaluation
true
source model id
MiniMax-M3
Original result ↗

OpenCompass LLM

88.3%

source id
openCompass
leaderboard name
Knowledge_Engineering
dimension
general
metric
Knowledge_Engineering
rank
8
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
KnowledgeTable
catalog
closed
snapshot
month
26-07
fileName
assets/llm-rank/llm-data-v2.26-07.20260824.json
updateTime
2026-08-24
columnFileName
assets/llm-rank/llm-column-v2.26-07.20260824.json
metric_kind
Knowledge_Engineering
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
1-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
15
benchmark_evaluation
true
source model id
MiniMax-M3
Original result ↗

Published prices

1–20 / 33

minimax/minimax-m3 ↗

$0.06

Inference provider
openRouter
Platform
OpenRouter
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions

minimax/minimax-m3 ↗

$0.30

Inference provider
openRouter
Platform
OpenRouter
Unit
input_tokens_per_million
Currency
USD
Billing conditions

minimax/minimax-m3 ↗

$1.20

Inference provider
openRouter
Platform
OpenRouter
Unit
output_tokens_per_million
Currency
USD
Billing conditions

minimax/minimax-m3:batch ↗

$0.04

Inference provider
openRouter
Platform
OpenRouter
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
mode
batch

minimax/minimax-m3:batch ↗

$0.22

Inference provider
openRouter
Platform
OpenRouter
Unit
input_tokens_per_million
Currency
USD
Billing conditions
mode
batch

minimax/minimax-m3:batch ↗

$0.88

Inference provider
openRouter
Platform
OpenRouter
Unit
output_tokens_per_million
Currency
USD
Billing conditions
mode
batch

MiniMaxAI/MiniMax-M3 ↗

$0.06

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
tier
0-512k-input
label
Input tokens within [0, 512k)
Billing label
Cached input tokens (per 1M)

MiniMaxAI/MiniMax-M3 ↗

$0.12

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
tier
over-512k-input
label
Input tokens over 512k
Billing label
Cached input tokens (per 1M)

MiniMaxAI/MiniMax-M3 ↗

$0.30

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
input_tokens_per_million
Currency
USD
Billing conditions
tier
0-512k-input
label
Input tokens within [0, 512k)
Billing label
Input tokens (per 1M)

MiniMaxAI/MiniMax-M3 ↗

$0.60

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
input_tokens_per_million
Currency
USD
Billing conditions
tier
over-512k-input
label
Input tokens over 512k
Billing label
Input tokens (per 1M)

MiniMaxAI/MiniMax-M3 ↗

$1.20

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
output_tokens_per_million
Currency
USD
Billing conditions
tier
0-512k-input
label
Input tokens within [0, 512k)
Billing label
Output tokens (per 1M)

MiniMaxAI/MiniMax-M3 ↗

$2.40

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
output_tokens_per_million
Currency
USD
Billing conditions
tier
over-512k-input
label
Input tokens over 512k
Billing label
Output tokens (per 1M)

minimax/minimax-m3 ↗

$0.06

Inference provider
vercelAIGatewayCatalog
Platform
fireworks
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
endpoint
fireworks | minimax/minimax-m3
model_type
language
billing_scope
via Vercel AI Gateway
Billing label
Cached input / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$0.30

Inference provider
vercelAIGatewayCatalog
Platform
fireworks
Unit
input_tokens_per_million
Currency
USD
Billing conditions
endpoint
fireworks | minimax/minimax-m3
model_type
language
billing_scope
via Vercel AI Gateway
Billing label
Input tokens / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$1.20

Inference provider
vercelAIGatewayCatalog
Platform
fireworks
Unit
output_tokens_per_million
Currency
USD
Billing conditions
endpoint
fireworks | minimax/minimax-m3
model_type
language
billing_scope
via Vercel AI Gateway
Billing label
Output tokens / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$0.10

Inference provider
vercelAIGatewayCatalog
Platform
gmicloud
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
endpoint
gmicloud | minimax/minimax-m3
model_type
language
quantization
fp8
billing_scope
via Vercel AI Gateway
context_tokens
min_inclusive
512000
Billing label
Cached input / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$0.05

Inference provider
vercelAIGatewayCatalog
Platform
gmicloud
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
endpoint
gmicloud | minimax/minimax-m3
model_type
language
quantization
fp8
billing_scope
via Vercel AI Gateway
context_tokens
max_exclusive
512000
min_inclusive
0
Billing label
Cached input / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$0.48

Inference provider
vercelAIGatewayCatalog
Platform
gmicloud
Unit
input_tokens_per_million
Currency
USD
Billing conditions
endpoint
gmicloud | minimax/minimax-m3
model_type
language
quantization
fp8
billing_scope
via Vercel AI Gateway
context_tokens
min_inclusive
512000
Billing label
Input tokens / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$0.24

Inference provider
vercelAIGatewayCatalog
Platform
gmicloud
Unit
input_tokens_per_million
Currency
USD
Billing conditions
endpoint
gmicloud | minimax/minimax-m3
model_type
language
quantization
fp8
billing_scope
via Vercel AI Gateway
context_tokens
max_exclusive
512000
min_inclusive
0
Billing label
Input tokens / 1M
Pricing via
Vercel AI Gateway

minimax/minimax-m3 ↗

$1.92

Inference provider
vercelAIGatewayCatalog
Platform
gmicloud
Unit
output_tokens_per_million
Currency
USD
Billing conditions
endpoint
gmicloud | minimax/minimax-m3
model_type
language
quantization
fp8
billing_scope
via Vercel AI Gateway
context_tokens
min_inclusive
512000
Billing label
Output tokens / 1M
Pricing via
Vercel AI Gateway

Provider specifications & access

1–3 / 3

MiniMax-M3 ↗

source model id
MiniMaxAI/MiniMax-M3
hosting
siliconflow-hosted
platforms
  • SiliconFlow
max output tokens
131072
architecture
input_modalities
  • text
  • image
output_modalities
  • text
capabilities
  • promptCaching
  • reasoning
  • structuredOutputs
  • toolCalling

MiniMax-M3 ↗

source model id
MiniMax-M3
hosting
minimax-hosted
platforms
  • MiniMax
context length
1000000
architecture
input_modalities
  • text
  • image
  • video
output_modalities
  • text
capabilities
  • videoInput
  • reasoning
  • toolCalling
  • streaming

MiniMax M3 ↗

source model id
minimax/minimax-m3
hosting
vercel-gateway
platforms
  • Vercel AI Gateway
context length
512000
max output tokens
512000
architecture
input_modalities
  • text
  • image
  • pdf
output_modalities
  • text
capabilities
  • promptCaching
  • reasoning
  • structuredOutputs
  • toolCalling
  • visionInput

Source specification observations

1–9 / 9

miniMax ↗

field
context_length
value
1000000
normalized value
1000000
scope
source-model
source id
miniMax
confidence
1

openRouter ↗

field
context_length
value
1048576
normalized value
1048576
scope
source-model
source id
openRouter
confidence
1

openRouter ↗

field
context_length
value
524288
normalized value
524288
scope
source-model
source id
openRouter
confidence
1

siliconFlow ↗

field
context_length
value
1049K
scope
source-model
source id
siliconFlow
confidence
1

vercelAIGatewayCatalog ↗

field
context_length
value
512000
normalized value
512000
scope
source-model
source id
vercelAIGatewayCatalog
confidence
1

openRouter ↗

field
max_output_tokens
value
512000
normalized value
512000
scope
source-model
source id
openRouter
confidence
1

openRouter ↗

field
max_output_tokens
value
512000
normalized value
512000
scope
source-model
source id
openRouter
confidence
1

siliconFlow ↗

field
max_output_tokens
value
131K
scope
source-model
source id
siliconFlow
confidence
1

vercelAIGatewayCatalog ↗

field
max_output_tokens
value
512000
normalized value
512000
scope
source-model
source id
vercelAIGatewayCatalog
confidence
1

Original sources

1–17 / 17

SiliconFlow · Model page ↗

Hugging Face · MiniMaxAI/Minimax-M3 ↗

Hugging Face · Model repository ↗

OpenRouter · Model page ↗

OpenRouter · Model page ↗

SiliconFlow · Playground ↗

Vercel AI Gateway · Model page ↗

MiniMax · Model documentation ↗

MiniMax · Model documentation ↗

SiliconFlow · API documentation ↗

Hugging Face · Catalog ↗

Hugging Face · Catalog ↗

MiniMax · Catalog ↗

OpenRouter · Catalog ↗

OpenRouter · Catalog ↗

SiliconFlow · Catalog ↗

Vercel AI Gateway · Catalog ↗

Source links

1–9 / 9

Explore more models

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace