Skip to main content

Beam 1.0 501B A23B

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...

Model card on Hugging Face
Input / output price
$0.25 / $1.00 per 1M
Context
1M tokens
Released
Oct 16, 2026
Agent setup

Paste into Claude Code, Cursor, or Codex

Help me set up the Respan gateway: fetch and follow https://www.respan.ai/docs/agent-setup.md. Use model "beam-1.0-501b-a23b".

Read docs

Providers

5 providers, cheapest first, OpenRouter last, priced per 1M tokens after gateway discounts.

ProviderInputOutput
Nebiusnebius/beam-1.0-501b-a23b
$0.25/M$1.00/M
IO.netionet/reflection/Beam-1.0-501B-A23B
$0.3375/M$1.10/M
Basetenbaseten/beam-1.0-501b-a23b
$0.40/M$1.20/M
TogetherAItogetherai/beam-1.0-501b-a23b
$0.40/M$1.20/M
OpenRouteropenrouter/beam-1.0-501b-a23b
$0.30/M$1.00/M

Is a provider down? OpenRouter status

Pricing

Averaged across 5 providers, in USD per 1M tokens after gateway discounts.

Average input
$0.34 per 1M
Average output
$1.10 per 1M
Average cache read
$0.07 per 1M

Cost by volume

Total cost from 0 to 1T tokens on the cheapest and most expensive provider.

Highest$0.40/MBaseten and TogetherAILowest$0.25/MNebius

Throughput and latency

30-day averages from requests through the Respan gateway. Under each figure, every model on the gateway by its average, this one in blue.

Output speed

Higher is better

45tokens/sacross providers

Faster than 45% of the other 354 models on the gateway

Response latency

Lower is better

0.5sacross providers

Faster than 87% of the other 412 models on the gateway

Benchmarks

The Respan Index plus 27 published scores from 4 sources, each linked to where it was published.

39.2/ 100

Respan Index#65 of 167

167 ranked models. The shaded band is this model's 95% interval, 29.6 to 56.2. Gemini 3 Pro Preview and Claude Sonnet 4.6 fall inside it, so the gap is within the margin of error.
Beam 1.0 501B A23B by category: Coding 16.9, Agents 55.8, Reasoning 38.7, Knowledge & language 35.4 (estimated), Human preference 52.0
  1. 65Beam 1.039.2
  2. 63Gemini 3 Pro Preview39.5
  3. 66Claude Sonnet 4.637.3

Published scores

BenchmarkScore
LiveCodeBenchhuggingface.co88.3%
SWEBench Verifiedhuggingface.co81.6%
Terminal Bench v2.1huggingface.co80.1%
SWEBench Multilingualhuggingface.co78.3%
SWE Bench Pro v2-Hardhuggingface.co76.5%
SWE Bench Pro v1huggingface.co66.1%
SciCodehuggingface.co49.2%
DeepSWE v1.1huggingface.co43.3%
SWE Atlas Codebase QnAhuggingface.co31.8%
Terminal-Bench 4.0Vals AI4%

Apps

Use Beam 1.0 501B A23B in the tools you already run. Each guide points the app at the Respan gateway.

Setup guides for more apps are in the integration docs.

Beam 1.0 501B A23B FAQ

Short answers, built from the data on this page.

Beam 1.0 501B A23B is a model from Reflection AI, released on Oct 16, 2026. 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...

On the Respan gateway, input starts at $0.25 per 1M tokens on Nebius, and output starts at $1.00 per 1M tokens on Nebius and OpenRouter, after gateway discounts. Averaged across its 5 providers, it costs $0.34 per 1M input tokens and $1.10 per 1M output tokens.

Beam 1.0 501B A23B is available through 5 providers: Nebius, IO.net, Baseten, TogetherAI, and OpenRouter.

Beam 1.0 501B A23B accepts up to 1M tokens of context on its largest provider route.

Point any OpenAI-compatible SDK at https://api.respan.ai/api/ with your Respan API key and set the model to "beam-1.0-501b-a23b". To set it up with a coding agent, copy the agent setup prompt at the top of this page.

Beam 1.0 501B A23B scores 39.2 out of 100 on the Respan Index v1.1, ranking #65 of 167 models. By category: Coding 16.9, Agents 55.8, Reasoning 38.7, Knowledge & language 35.4 (estimated), and Human preference 52.0. The Respan Index combines independent and lab-reported benchmark results into one score, using benchmark data collected by Respan.

Beam 1.0 501B A23B has 27 published benchmark scores, including LiveCodeBench 88.3%, MCP Atlas 80.8%, and GPQA Diamond 90.6%. Each score links to where it was published; the data is collected by Respan.

Certified where it counts.

Built to meet the security and privacy standards that enterprise and healthcare teams require.

ISO 27001

The internationally recognized standard for information security management.

SOC 2

Secure, compliant management of your data across all of our systems.

GDPR

Operated under GDPR, the world's strictest standard for data privacy.

HIPAA

HIPAA compliant, with a BAA available for healthcare teams.

Works with your entire stack

Use Respan with your favorite frameworks and tools.

AnthropicAI21 LabsxAIVercel AI SDK
LangChainOpenRouterAssemblyAICursor
LiteLLMGroqBasetenMem0
CohereCogneeLlamaIndexFireworks
DeepSeekLinkupHaystackTogether AI
PerplexityInference.netPostHogRubyLLM
SlackMastraAzure OpenAIMistral
AWS BedrockMoonshotHyperspellBAML
ModaInstructorGoogle GeminiParasail
ReductoZapierAgnoNebius
OpenAINovita AIReplicateSuperagent

Built for AI agents.
Break less.
Ship more.