# Beam 1.0 501B A23B Usage, Cost & Rank | OpenCode Data

> Beam 1.0 501B A23B costs $0.00 per 1M input tokens and $0.00 per 1M output tokens.

Sparse mixture-of-experts reasoning model with 23B active parameters for coding and agentic workflows

- Page: https://opencode.ai/data/reflection/beam-1-0-501b-a23b
- JSON: https://opencode.ai/data/reflection/beam-1-0-501b-a23b.json
- Model ID: reflection/beam-1.0-501b-a23b
- Lab: Reflection

## Model facts

| Fact | Value |
| --- | --- |
| Context window | 1M |
| Max output | 131K |
| Knowledge cutoff | 2026-06-30 |
| Release date | 2026-10-05 |
| Input modalities | text |
| Output modalities | text |
| Reasoning | Yes |
| Tool calling | Yes |
| Open weights | Yes |
| Weights | [Hugging Face](https://huggingface.co/reflection/Beam-1.0-501B-A23B) |
| Weights | [Hugging Face (FP8)](https://huggingface.co/reflection/Beam-1.0-501B-A23B-FP8) |
| Weights | [Hugging Face (NVFP4)](https://huggingface.co/reflection/Beam-1.0-501B-A23B-NVFP4) |

## Pricing: USD per 1M tokens

| Input | Output | Cached input | Cache write |
| --- | --- | --- | --- |
| $0.00 | $0.00 | $0.00 | - |

## Benchmarks

| Benchmark | Score | Metric | Source |
| --- | --- | --- | --- |
| DeepSWE | 43.3 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SWE-Bench Pro | 76.5 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SWE-Bench Pro | 66.1 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| Terminal-Bench | 80.1 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SWE-Atlas Codebase QnA | 31.8 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SWE-Bench Multilingual | 78.3 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SWE-Bench Verified | 81.6 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| AIME 2026 | 97.7 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| Humanity's Last Exam | 36 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| Humanity's Last Exam | 51.1 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| SciCode | 49.2 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| CritPt | 16.3 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| GPQA Diamond | 90.6 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| LiveCodeBench | 88.3 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| AutomationBench | 39.3 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| MCP Atlas | 80.8 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| τ3 Banking | 36.1 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| BrowseComp | 76.8 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| DeepSearchQA | 80.5 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| AA-LCR | 78.2 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| LongBench | 67.5 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| IFBench | 79.6 |  | https://huggingface.co/reflection/Beam-1.0-501B-A23B |
| AA-Omniscience | 10.8 | index | https://huggingface.co/reflection/Beam-1.0-501B-A23B |

## Methodology

- Updates: Aggregated every hour. Days and weeks use UTC.
- Tokens: Input, output, reasoning, and cached tokens for each request.
- Users and sessions: Approximate counts of distinct users and OpenCode sessions.
- Cost: Session cost is the average cost per OpenCode session. Token prices are list prices from the OpenCode model catalog.
- Retention: The share of a model's users in one week who use it again the next week.
- Citation: Cite OpenCode Data (opencode.ai/data) with the update time shown at the top of the page.