reflection
Reflection: Beam 1.0 501B A23B
reflection/beam-1.0-501b-a23b
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Beam 1.0 501B A23B (reflection/beam-1.0-501b-a23b) is Reflection's open-weight, Apache-2.0 instruction-tuned reasoning model: a sparse mixture-of-experts with 501B total and about 23B active parameters per token, a 1M-token context window, and text-only input and output. The vendor positions it for agentic coding, MCP-based tool use, and math and science reasoning, and reports strong benchmark results plus token-efficient reasoning. Reasoning is mandatory with tunable effort levels. It was developed predominantly for English; validate it on your own workloads.[1][2]
Strengths
- Vendor-reported strengths in agentic software engineering: resolving real GitHub issues, CLI and DevOps work, and integration with agentic coding harnesses, backed by a vendor-published benchmark table.[2]
- Sparse MoE design with 501B total and about 23B active parameters, a 1M-token context window, and open Apache-2.0 weights released in BF16, FP8, and NVFP4 numerics.[1][2]
- Always-on reasoning with selectable effort levels from low to max; the vendor claims comparable solution quality using substantially fewer inference-time reasoning tokens.[1][2]
Limitations
- Text-only input and output, and the model card states it was developed predominantly for English; behavior in other languages is undocumented.[1][2]
- Most capability and benchmark evidence is vendor-reported; the OpenRouter catalog adds some arena and index data, but independent verification is thin, so validate on your own workloads.[1][2]
- Reasoning cannot be disabled according to the OpenRouter catalog, which may be wasteful for simple prompts, and the technical report was still marked as coming.[1][2]
Where to use it
- Agentic coding assistants and CI or DevOps automation that resolve issues or execute multi-step terminal workflows.[2]
- Long-context work over large codebases or documents that can exploit the 1M-token window.[1][2]
- Tool-calling agents using MCP integration, browsing, and structured outputs for multi-step tasks driven by natural-language instructions.[1][2]
Sources
- [1] OpenRouter catalog · Reflection: Beam 1.0 501B A23BSources checked: Nov 13, 2026
- [2] Model documentation · Reflection: Beam 1.0 501B A23BSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
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...
Specifications
- Model ID
- reflection/beam-1.0-501b-a23b
- Canonical ID
- reflection/beam-1.0-501b-a23b-20261016
- Creator
- reflection
- Added to OpenRouter
- 16.10.2026
- Context length
- 1,048,576
- Modality
- text->text
- Input modalities
- text
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.300 / out $1
- Knowledge cutoff
- 30.06.2026
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | CC — below the B band | reasoning: mandatory · default max · efforts: max, xhigh, high, medium, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · default max · efforts: max, xhigh, high, medium, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Document analysis | No measurement | — |
| Vision | No measurement | — |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 17.7 | CC — below the B band | 13.11.2026 |
| Design Arena | models · 3dwin rate 33% · rank 80 | 1,123Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 44.3% · rank 51 | 1,227Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 40.1% · rank 61 | 1,187Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 48.6% · rank 38 | 1,259Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 46.8% · rank 46 | 1,245Elo | BB — next 30% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Nebiusfp4 | 1,048,576 | $0.250 | $1 | — | — | 100% |
| CoreWeavenvfp4 | 1,048,576 | $0.300 | $1 | — | — | 100% |
| DigitalOcean | 1,048,576 | $0.400 | $1.2 | — | — | 100% |
| BaseTenfp4 | 1,048,576 | $0.400 | $1.2 | — | — | 100% |
| Together | 1,048,576 | $0.400 | $1.2 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.