Anthropic · Frontier
Claude Sonnet 4.6 Thinking
Claude Sonnet 4.6 with extended thinking enabled for tougher coding, planning, and multi‑tool tasks. Ideal for long‑horizon agent workflows and complex problem solving.
- Reasoning
- Tools
- Vision
- Structured output
- PDF input
- Input
- Text, Images
- Output
- Text
- Reasoning effort
- low · medium · high · xhigh · max
- Discovery tier
- Frontier
Pricing
USD per 1M tokens- Input
- $3.00
- Output
- $15.00
- Cache read
- $0.30
Reference prices as published by the model provider. API example
Specifications
- Context window
- 1M
- Max output
- 128K
- Average output (7 days)
- 598 tokens
- Longer than 56% of 473 models
Side by side
Claude Sonnet 4.6 Thinking compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| Claude Sonnet 4.6 ThinkingAnthropicThis model | GPT Terra LatestOpenAIRemove | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Discovery tier | Frontier | Frontier |
| Added | Feb 17, 2026 | Sep 11, 2026 |
| Limits | ||
| Context window | 1M | 1.05MBest |
| Max output | 128K | 128K |
| Average output | 598 tokens | 223 tokens |
| Pricing, USD per 1M tokens | ||
| Input | $3.00 | $2.00Best |
| Output | $15.00 | $12.00Best |
| Cache read | $0.30 | $0.20Best |
| Cache write | — | $2.50 |
| Performance | ||
| Throughput | 97.5 tok/s | — |
| Time to first token | 3.1 s | — |
| Request success (24h) | 100.0% | 100.0% |
| Benchmark indexes | ||
| Intelligence Index | 30.1 | 30.1 |
| Coding Index | 63.0 | 64.7Best |
| Capabilities | ||
| Reasoning | Yes | Yes |
| Tools | Yes | Yes |
| Vision | Yes | Yes |
| Structured output | Yes | Yes |
| PDF input | Yes | Yes |
| Reasoning effort | low, medium, high, xhigh, max | none, low, medium, high, xhigh, max |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- Claude Sonnet 4.6 Thinking
- GPT Terra Latest
Intelligence Index
- Claude Sonnet 4.6 Thinking30.1
- GPT Terra Latest30.1
Claude Sonnet 4.6 Thinking tested at max reasoning effort
Claude Sonnet 4.6 Thinking: Better than 73% of 326 scored models
Indexes
Coding Index
Composite of the coding evaluations
- Claude Sonnet 4.6 Thinking: 63.0
- GPT Terra Latest: 64.7Top
Agentic work
AutomationBench-AA
Workflow automation, penalized for guardrail violations
- Claude Sonnet 4.6 Thinking: 20.1%
- GPT Terra Latest: 33.5%Top
AutomationBench-AA tasks completed
Workflows finished end to end without a guardrail violation
- Claude Sonnet 4.6 Thinking: 3.7%
- GPT Terra Latest: 7.2%Top
AA-Briefcasev1.1
Agentic knowledge work, rated head to head
- Claude Sonnet 4.6 Thinking: 1,062 EloTop
- GPT Terra Latest: 1,038 Elo
GDPval-AAv2.1
Economically valuable occupational tasks, rated head to head
- Claude Sonnet 4.6 Thinking: 1,233 Elo
- GPT Terra Latest: 1,266 EloTop
Document reasoning
GDP.pdf
Professional reasoning over PDF documents, all-pass rate
- Claude Sonnet 4.6 Thinking: 15.8%
- GPT Terra Latest: 17.0%Top
AA-LCRv1.1
Reasoning across long documents
- Claude Sonnet 4.6 Thinking: 80.0%Top
- GPT Terra Latest: 74.0%
MLCR-AA
Long-context reasoning over medical records
- Claude Sonnet 4.6 Thinking: 7.8%
- GPT Terra Latest: 22.8%Top
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- Claude Sonnet 4.6 Thinking: 33.6%Top
- GPT Terra Latest: 33.3%
GPQA Diamond
Graduate-level science questions
- Claude Sonnet 4.6 Thinking: 87.5%Top
- GPT Terra Latest: 87.2%
IFBench
Following precise output instructions
- Claude Sonnet 4.6 Thinking: 56.6%
- GPT Terra Latest: 62.2%Top
Coding
Terminal-Benchv4.0
Practical software engineering in a terminal
- Claude Sonnet 4.6 Thinking: 3.0%Top
- GPT Terra Latest: 1.0%
SciCode
Scientific computing in Python
- Claude Sonnet 4.6 Thinking: 50.1%
- GPT Terra Latest: 50.5%Top
Legacy evaluations
Terminal-Bench Hard
Earlier terminal task suite
- Claude Sonnet 4.6 Thinking: 53.0%
- GPT Terra Latest: —
τ²-Bench Telecom
Tool use in a customer-support simulation
- Claude Sonnet 4.6 Thinking: 75.7%Top
- GPT Terra Latest: 72.8%
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- Claude Sonnet 4.6 Thinking: 80.0%Top
- GPT Terra Latest: 74.0%
Scores from Artificial Analysis, updated Oct 26, 2026. Claude Sonnet 4.6 Thinking was tested at max reasoning effort; compare scores measured at the same effort.Artificial Analysis
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Throughput
- 97.5 tok/s
- Faster than 61% of 256 models
- Time to first token
- 3.1 s
- Quicker than 50% of 256 models
- Prompt cache hits
- 83.6%
- About $0.74 per 1M input tokens with caching
- Request success (24h)
- 100.0%
- Reliable
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-sonnet-4.6:thinking",
"reasoning_effort": "medium",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="anthropic/claude-sonnet-4.6:thinking",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)Anthropic
Related models
Other Anthropic models, starting with the same family.