Inference
EN
Reflection AI

Beam 1.0 501B-A23B NVFP4

Beam 1.0 501B-A23B NVFP4 is a reasoning model from Reflection AI in the Beam family. 17 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-10
RankedEnough results for an index.
IndexAll categories, out of 100.58.5±5.4Median 51.0
CoverageIndex weight that has results.85%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.1MMedian 262K
EloLMArena rating and rank.—Median 1422

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.7 counted55.0103 / 218
CodingCode writing and repair.3 counted60.866 / 209
ReasoningLogic problems and puzzles.3 counted62.755 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.2 counted59.269 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted63.0—
MathMath problems.1 counted55.5—

Results

17 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
AIME 2026Math97.5%4 / 34
Software Engineering Benchmark VerifiedCoding80.7%15 / 114
LongBench v2Reasoning67.2%4 / 25
SWE-bench ProCoding65.6%15 / 83
Critical Physics TasksReasoning15.4%48 / 214
Artificial Analysis Long Context ReasoningReasoning78.7%61 / 217
MCP AtlasAgentic80.5%15 / 52
Instruction Following BenchmarkInstruction79.2%17 / 57
Scientific Code BenchmarkCoding46.9%14 / 35
GPQA Diamond · BenchLMKnowledge89.3%36 / 84
Terminal-Bench 2.1 · BenchLMAgentic78.4%40 / 86
DeepSearchQAAgentic81.1%13 / 24
BrowseCompAgentic75.6%38 / 63
Humanity's Last Exam without toolsKnowledge34.8%33 / 53
τ³-Bench Tool-Agent-User EvaluationAgentic35.4%22 / 32
Humanity's Last Exam with toolsAgentic51.1%25 / 36
DeepSWEAgentic42.7%56 / 59
SWE-Atlas Codebase QnACoding31.4%—

Sources

BenchLMCC BY-NC 4.0

History

5 releases
Index59.5 → 58.5
23 Oct 202612 Nov 2026

Questions

How good is Beam 1.0 501B-A23B NVFP4?
Its index is 58.5 out of 100. That is place 82 of 283 ranked models.
What is Beam 1.0 501B-A23B NVFP4 best at?
Reasoning: 62.7 out of 100, place 55 of 227.
What is the context window of Beam 1.0 501B-A23B NVFP4?
1M tokens.
Is Beam 1.0 501B-A23B NVFP4 open source?
Yes. Its weights are published.
When was Beam 1.0 501B-A23B NVFP4 released?
October 2026.

Badge

Beam 1.0 501B-A23B NVFP4 on Inference360: place 82 of 283, index 58.5
HTML
<a href="https://inference360.ai/models/reflection-beam-nvfp4"><img src="https://inference360.ai/badges/reflection-beam-nvfp4.svg" alt="Beam 1.0 501B-A23B NVFP4 on Inference360: place 82 of 283, index 58.5" height="20"></a>
Markdown
[![Beam 1.0 501B-A23B NVFP4 on Inference360: place 82 of 283, index 58.5](https://inference360.ai/badges/reflection-beam-nvfp4.svg)](https://inference360.ai/models/reflection-beam-nvfp4)

Compare with

Kimi K2.658.2Apodex 1.158.9Claude Opus 4.658.0Beam 1.0 501B-A23B FP859.2

More from Reflection AI

All models
Beam 1.0 501B-A23B60.2Beam 1.0 501B-A23B FP859.2