Reflection 5 indexed models
5 matching models
reflection/beam-1.0-501b-a23b-20261016
Beam 1.0 is Reflection's open-weight reasoning model for coding and long-horizon agentic workflows. It uses a sparse mixture-of-experts architecture with 501B total and 23B active parameters per token, and a 1M-token context window. Beam targets software engineering, multi-step
Context window 1,048,576
Input modalities text
Output modalities text
Capabilities promptCaching, reasoning, structuredOutputs, toolCalling Benchmarks Artificial Analysis: 17.7 LMArena: 1,424.689 OpenRouter Benchmarks: 89.394% Inference prices Published price : $0.04 / cache_read_tokens_per_million · openRouter ↗ Published price : $0.30 / input_tokens_per_million · openRouter ↗ Published price : $1.00 / output_tokens_per_million · openRouter ↗ vercelAIGatewayCatalog/reflection/beam
Beam is Reflection’s reasoning model for coding and agentic workflows, combining strong coding capabilities with fast, low-latency inference.
Context window 1,048,576
Maximum output 131,072
Input modalities text
Output modalities text
Capabilities promptCaching, reasoning, structuredOutputs, toolCalling Benchmarks Vercel AI Gateway: 23.352% Inference prices cheaperInference/beam-free
Context window 1,048,576
Maximum output 131,072
Input modalities text
Output modalities text
Capabilities promptCaching, reasoning, structuredOutputs, toolCalling Inference prices azureFoundry/azure-huggingface/reflection--beam-1.0-501b-a23b-fp8@1
FP8-quantized Beam-1.0 501B-A23B mixture-of-experts language model for text chat.
Context window 1,048,576
Input modalities text
Output modalities text azureFoundry/azure-huggingface/reflection--beam-1.0-501b-a23b-nvfp4@1
NVFP4-quantized Beam-1.0 501B-A23B mixture-of-experts language model for text chat.
Context window 1,048,576
Input modalities text
Output modalities text
EVIDENCE / COMPARISON
Benchmarks
Explore every benchmark
Loading
Search evaluations
Dimension
All dimensions
Source
All sources
Entries
Top 12
Top 25
Top 50
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.