someaistuff.
reflectionReleased Oct 16, 2026

Reflection: Beam 1.0 501B A23B

Apache-2.0 text model with 501B total and 23B active parameters; ships BF16, FP8 and NVFP4 weights.

Source description

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...

Compare
Context window1M
Max output943.7K
Parameters501B
Provider endpoints5
Benchmark results19
Tool callingStructured outputReasoningPrompt cachingOpen weights
Measured performance

Benchmark results

All leaderboards ↗
AA-Briefcase ↗1.1
976.2

Effort not specified

Independent result
GDPval-AA ↗2.1
1,177.08

Effort not specified

Independent result
AutomationBench-AA ↗AA Index 4.3.2 collection
48.46%

Effort not specified

Independent result
SciCode ↗AA Index 4.3.2 collection
48.5%AA · Under review

Effort not specified

Independent result
GDP.pdf ↗AA Index 4.3.2 collection
11.6%

Effort not specified

Independent result
CritPt ↗AA Index 4.3.2 collection
13.14%AA · Under review

Effort not specified

Independent result
AA-Omniscience ↗AA Index 4.3.2 collection
1.5

Effort not specified

Independent result
AA-LCR ↗1.1
78%

Effort not specified

Independent result
LiveBench Overall ↗2.0 · 2026-10-14 (dataset 2026-09-02)
55.5%

Effort: max · single run · equal mean of 7 suites

Independent result
LiveBench Design ↗2.0 · 2026-10-14 (dataset 2026-09-02)
67.9%

Effort: max · single run

Independent result
LiveBench Research ↗2.0 · 2026-10-14 (dataset 2026-09-02)
72.8%

Effort: max · single run

Independent result
LiveBench Browser Use ↗2.0 · 2026-10-14 (dataset 2026-09-02)
0%

Effort: max · single run

Independent result
LiveBench Coding ↗2.0 · 2026-10-14 (dataset 2026-09-02)
83.9%

Effort: max · single run

Independent result
LiveBench Agentic ↗2.0 · 2026-10-14 (dataset 2026-09-02)
17.3%

Effort: max · single run

Independent result
LiveBench Reasoning ↗2.0 · 2026-10-14 (dataset 2026-09-02)
80.2%

Effort: max · single run

Independent result
Access & cost

Prices across providers

5 quotes

Your token mix

USD · listed token rates

Nebius · via OpenRouternebius/fp8 · quantization=fp8
$0.50
$0.70
CoreWeave · via OpenRoutercoreweave/fp4 · quantization=fp4
$1.75
BaseTen · via OpenRouterbaseten/fp4 · quantization=fp4
$2.375

OpenRouter

Catalog quote
Input
$1.00
Output
$3.00
Cached input
$0.10
USD / 1M tokens
1M context943.7K max outputPrice source ↗Checked Oct 17, 2026
Endpoint IDreflection/beam-1.0-501b-a23b

Nebius

via OpenRouter

nebius/fp8 · quantization=fp8

Input
$0.25
Output
$1.00
USD / 1M tokens
1M context131.1K max outputPrice source ↗Checked Oct 17, 2026
Endpoint IDreflection/beam-1.0-501b-a23b::nebius/fp8

Together

via OpenRouter

together

Input
$0.40
Output
$1.20
Cached input
$0.04
USD / 1M tokens
1M context131.1K max outputPrice source ↗Checked Oct 17, 2026
Endpoint IDreflection/beam-1.0-501b-a23b::together

CoreWeave

via OpenRouter

coreweave/fp4 · quantization=fp4

Input
$1.00
Output
$3.00
Cached input
$0.10
USD / 1M tokens
1M context943.7K max outputPrice source ↗Checked Oct 17, 2026
Endpoint IDreflection/beam-1.0-501b-a23b::coreweave/fp4

BaseTen

via OpenRouter

baseten/fp4 · quantization=fp4

Input
$1.30
Output
$4.30
Cached input
$0.13
USD / 1M tokens
1M context131.1K max outputPrice source ↗Checked Oct 17, 2026
Endpoint IDreflection/beam-1.0-501b-a23b::baseten/fp4

Model specifications

Developer
reflection
Modality
Language
Context
1,048,576 tokens
Maximum output
943,718 tokens
Weights
Open weights
Released
Oct 16, 2026
Provider listed
Oct 16, 2026
First discovered
Oct 16, 2026

Exact identities

  • openrouterreflection/beam-1.0-501b-a23b
  • artificial-analysis69468cce-7b5c-414a-b498-41df4c34df85
  • livebenchbeam-1.0

Sources

OpenRouter ↗Artificial Analysis ↗LiveBench ↗Model documentation ↗

Updated Oct 17, 2026