LLM Reference

Claude Opus 4.6

Released
2026-02-05
Last refreshed
2026-06-29
Status
Researched 59d ago
ProprietaryCommercial use: conditionalMultimodalCodingRAGAgentsLong contextVisionClassificationJSON / Tool useHighlight

Claude Opus 4.6 is worth evaluating for coding, rag, and agents when its provider route and context window match the workload.

Use it for

  • Teams evaluating coding, rag, and agents
  • Workloads that can use a 1m context window
  • Buyers comparing 4 tracked provider routes

Do not use it for

  • Workloads where another current model has stronger sourced task evidence
Specifications
Released
2026-02-05
Context
1m
Architecture
Decoder Only
Knowledge cutoff
2025-12
Specialization
general
Openness
Proprietary
License
ProprietaryCommercial use: conditional
Weights
Not released
Code
Unknown
Training
Fine-tuned
Created by
AnthropicAnthropic

Developing safe and ethical AI systems.

San Francisco, California, United States
Founded 2021
Website
Pricing
Output / 1M
$25.00
Input / 1M
$5.00

Cheapest of 6 routes · Anthropic · cache read $0.500

Providers(6)
AnthropicAWS BedrockGCP Vertex AIMicrosoft FoundryOpenRouterVercel AI Gateway
View 6 provider routes
Links
Website

About

Claude Opus 4.6 is Anthropic's Claude 4.6 model with multimodal text and image input and an optional reasoning mode. It offers a 1M-token context window and scores 80.8 on SWE-bench Verified.

Claude Opus 4.6 is a proprietary model in the Claude 4.6 family. The structured metadata tracks a 1m-token context window, multimodal input, reasoning, function calling, tool use, structured outputs, and code execution. This page tracks provider routes through Anthropic, AWS Bedrock, GCP Vertex AI, and 3 more, with the cheapest tracked route listed at $5 input and $25 output per 1M tokens. Headline tracked benchmarks include SWE-bench Verified 80.8, Google-Proof Q&A 91.3, and SWE-bench Pro 53.4.

Top use-case fit: coding, agents, and build tasks

Coding

Q/$ D

5 relevant benchmarks in the decision map.

RAG

Included by capability and metadata signals in the decision map.

Agents

Q/$ D

2 relevant benchmarks in the decision map.

Provider price ladder

Compare all 6

Compare API pricing across 4 providers for input and output tokens, batch, and cached reads when available.

ProviderInput / 1MOutput / 1MBatch in / outCacheRoute
Anthropic$5.00$25.00$2.50 / $12.50read $0.500 / 5m $6.25 / 1h $10.00
Serverless
Microsoft Foundry$5.00$25.00-read $0.500 / 5m $6.25 / 1h $10.00
Serverless
OpenRouter$5.00$25.00-read $0.500 / 5m $6.25
Serverless
Vercel AI Gateway$5.00$25.00-read $0.500
Serverless

Available via routers & gateways(16)

LiteLLM

Gateway

Open-source Python SDK and proxy server that unifies 100+ LLM APIs behind a single OpenAI-compatible interface, with load balancing, cost tracking, and configurable failover.

Free OSSAnthropicGCP Vertex AIMicrosoft Foundry

OpenRouter

Hybrid

Unified hybrid gateway to 400+ models from 60+ providers via a single OpenAI-compatible API, with optional auto-routing that selects the best model per prompt.

PassthroughAnthropicGCP Vertex AI

Portkey

Gateway

Production AI gateway routing to 1,600+ LLMs with failover, load balancing, semantic caching, and guardrails; Apache 2.0 core is fully self-hostable with the complete feature set.

SubscriptionAnthropicGCP Vertex AIMicrosoft Foundry

AIRouter

Router

Commercial LLM router that analyzes incoming requests and routes to the optimal model for cost/quality/latency via a drop-in OpenAI-compatible API, with a privacy-preserving embedding mode that avoids sending prompt content.

Passthrough + feeAnthropicGCP Vertex AI

Amazon Bedrock Intelligent Prompt Routing

Router

AWS Bedrock's native intelligent prompt router that routes prompts between Anthropic Claude model tiers (Haiku/Sonnet) based on predicted task complexity, with no extra per-routing charge.

PassthroughAWS Bedrock

Azure AI Foundry Model Router

Router

Microsoft Azure AI Foundry's native model router that uses a trained ML model to route each prompt in real time to the optimal Azure-hosted model, with Balanced/Cost/Quality mode selection and automatic failover.

PassthroughMicrosoft Foundry

Capabilities

VisionMultimodalReasoningFunction CallingTool UseStructured OutputsCode Execution

Benchmark peer barsfor Coding

SWE-bench ProRank 30 of 41
Claude Fable 5
80.3
Claude Opus 4.8
69.2
Grok 4.5
64.7
GPT-5.6 Sol
64.6
Claude Opus 4.6current
53.4
SWE-bench VerifiedRank 10 of 80
Claude Fable 5
96.0
Claude Mythos Preview
93.9
Claude Opus 4.8
88.6
Claude Opus 4.7
87.6
Claude Opus 4.6current
80.8
HumanEvalRank 3 of 97
Claude Sonnet 4.6
98.0
o3
96.7
Claude Opus 4.6current
95.0
Grok-3
94.5
GPT-5.5
94.2

Benchmark scores(20)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionSource
SWE-bench Verified80.82026-02https://www.anthropic.com/news/claude-4
Google-Proof Q&A91.3diamondhttps://www.anthropic.com/claude/opus
SWE-bench Pro53.4—https://anthropic.com/glasswing
MMLU PRO89.1—https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro
τ-bench84.8τ-benchhttps://benchlm.ai/benchmarks/tauBench
Chatbot Arena1501.0Thinkinghttps://arena.ai/leaderboard
MMMU Pro77.3official Anthropic system card, adaptive thinking, max effort, with image cropping tool, avg 5 runshttps://www.anthropic.com/claude-opus-4-6-system-card
SWE-rebench65.3pass@1 (best of 5 runs)https://swe-rebench.com/leaderboard
Aider Polyglot72.0Listed as 'Claude Opus 4 (32k thinking)' = 72 (percent_correct)https://aider.chat/docs/leaderboards/
AIME 202594.2AIME 2025 (accuracy)https://automatio.ai/models/claude-opus-4-6
ARC-AGI-268.8ARC-AGI-2 (accuracy%)https://llm-stats.com/benchmarks/arc-agi-v2
BrowseComp84.0BrowseComp (accuracy)https://www.vellum.ai/blog/claude-opus-4-6-benchmarks
Humanity's Last Exam53.0HLE with tools enabled (accuracy)https://www.anthropic.com/news/claude-opus-4-6
HumanEval95.0HumanEval (pass@1)https://automatio.ai/models/claude-opus-4-6
LiveCodeBench70.2LiveCodeBench score (accuracy)https://automatio.ai/models/claude-opus-4-6
MathVista74.8MathVista accuracy (accuracy)https://automatio.ai/models/claude-opus-4-6
MCP-Atlas62.7MCP-Atlas (accuracy%)https://llm-stats.com/benchmarks/mcp-atlas
Massive Multitask Language Understanding91.1MMLU (accuracy)https://automatio.ai/models/claude-opus-4-6
Massive Multi-discipline Multimodal Understanding76.5MMMU (accuracy)https://automatio.ai/models/claude-opus-4-6
Terminal-Bench 2.065.4Terminal-Bench 2.0 (accuracy%)https://llm-stats.com/benchmarks/terminal-bench-2

Migration checks

Coming from Claude 3 Opus?

Compare Claude Opus 4.6 with other models

Show all 75 popular comparisonssorted by 7-day search impressions
Claude Opus 4.6 vs Hunyuan Hy3 Preview534Claude Opus 4.6 vs DeepSeek V4 Flash429Claude Opus 4.6 vs Qwen3.6-27B414Claude Opus 4.6 vs Claude 3.7 Sonnet378Claude Opus 4.6 vs GPT-5.5338Claude Opus 4.6 vs Xiaomi MiMo-V2.5308Claude Opus 4.6 vs Qwen3-Max289Claude Opus 4.6 vs Qwen3.6-35B-A3B260Claude Opus 4.6 vs Qwen3-235B-A22B213Claude Opus 4.6 vs Step 3.5 Flash205Claude Opus 4.6 vs Xiaomi MiMo-V2.5-Pro190Claude Opus 4.6 vs Kimi K2.6183Claude Opus 4.6 vs Llama 3 8B Instruct144Claude Opus 4.6 vs GLM-5130Claude Opus 4.6 vs Claude Sonnet 4.5119Claude Opus 4.6 vs StepFun Step-2114Claude Opus 4.6 vs Qwen3.5-27B110Claude Opus 4.6 vs Gemini 2.5 Pro102Claude Opus 4.6 vs Qwen3.5-397B-A17B94Claude Opus 4.6 vs DeepSeek R1 Distill Llama 70B76Claude Opus 4.6 vs GPT-5.5 Pro75Claude Opus 4.6 vs Kimi K2.572Claude Opus 4.6 vs o3 Mini69Claude Opus 4.6 vs Llama 3 70B Instruct68Claude Opus 4.6 vs GPT-5.2 Codex68Claude Opus 4.6 vs Trinity-Large-Thinking67Claude Opus 4.6 vs DeepSeek R1 052867Claude Opus 4.6 vs DeepSeek V3.264Claude Opus 4.6 vs Qwen3.5-35B-A3B58Claude Opus 4.6 vs Qwen3.6 Max Preview58Claude Opus 4.6 vs Ling-2.6-1T54Claude Opus 4.6 vs DeepSeek V349Claude Opus 4.6 vs Llama 3.1 70B Instruct43Claude Opus 4.6 vs Qwen2.5-72B-Instruct36Claude Opus 4.6 vs Kimi K2 Thinking34Claude Opus 4.6 vs Claude Opus 4.534Claude Opus 4.6 vs GLM-5V-Turbo33Claude Opus 4.6 vs Tencent Hunyuan Turbo S31Claude Opus 4.6 vs Grok-330Claude Opus 4.6 vs Qwen2.5-7B-Instruct29Claude Opus 4.6 vs GPT-5.429Claude Opus 4.6 vs Qwen2.5-72B26Claude Opus 4.6 vs Qwen2.5-Max26Claude Opus 4.6 vs Llama Guard 3 1B24Claude Opus 4.6 vs Mistral Large 222Claude Opus 4.6 vs Claude Mythos Preview21Claude Opus 4.6 vs o3 Deep Research19Claude Opus 4.6 vs Gemini 2.5 Flash18Claude Opus 4.6 vs DeepSeek V3.118Claude Opus 4.6 vs GLM-5 Turbo16Claude Opus 4.6 vs Kimi K2 Thinking Turbo15Claude Opus 4.6 vs Llama 2 13B Chat15Claude Opus 4.6 vs Code Davinci 00115Claude Opus 4.6 vs Mistral Nemotron15Claude Opus 4.6 vs Gemma 7B Instruct14Claude Opus 4.6 vs Phi-3 Mini 4k13Claude Opus 4.6 vs Together AI - Llama 3 8B Lite13Claude Opus 4.6 vs Llama Guard 4 12B11Claude Opus 4.6 vs Llama 3.2 1B Instruct10Claude Opus 4.6 vs GPT-4o (2024-11-20)10Claude Opus 4.6 vs Qwen3.5-122B-A10B9Claude Opus 4.6 vs GPT-5.4-Cyber9Claude Opus 4.6 vs Trinity-Large-Preview7Claude Opus 4.6 vs Gemma 2 9B SahabatAI Instruct7Claude Opus 4.6 vs Kimi K2 Instruct6Claude Opus 4.6 vs Phi-4 Mini Flash Reasoning5Claude Opus 4.6 vs Qwen3-9B5Claude Opus 4.6 vs Mixtral 8x22B Instruct v0.34Claude Opus 4.6 vs DeepSeek R1 Zero4Claude Opus 4.6 vs Qwen2-7B-Instruct4Claude Opus 4.6 vs ShieldGemma 9B3Claude Opus 4.6 vs Mixtral 8x7B2Claude Opus 4.6 vs Ling-2.6-Flash1Claude Opus 4.6 vs GPT-4o Search Preview1Claude Opus 4.6 vs GLM-5 9B0

Frequently asked questions

What is the context window of Claude Opus 4.6?

Claude Opus 4.6 has a context window of 1m tokens.

How much does Claude Opus 4.6 cost?

Claude Opus 4.6 pricing ranges from $5.00/1M to $15/1M input tokens depending on the provider.

When was Claude Opus 4.6 released?

Claude Opus 4.6 was released on 2026-02-05.

Which providers offer Claude Opus 4.6?

Claude Opus 4.6 is available from 6 providers: Anthropic, AWS Bedrock, GCP Vertex AI, Microsoft Foundry, OpenRouter, Vercel AI Gateway.

What benchmarks has Claude Opus 4.6 been tested on?

Claude Opus 4.6 has been evaluated on 20 benchmarks, including SWE-bench Verified, Google-Proof Q&A, SWE-bench Pro, MMLU PRO, τ-bench.