Anthropic

Claude Opus 4.8

Tool callingStructured outputReasoningImage inputFile input
Claude Opus 4.8 launch post cover

Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents

Context
1M
Released
4mo ago
Input / 1M tokens
$5.00
Output / 1M tokens
$25.00
Cached input / 1M tokens
$0.50
Cache write / 1M tokens
$6.25

Sources:models.devOpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

DeepSWE 1.1

Primary coding-agent benchmark · 113 tasks · Data generated 2026-09-22

Official leaderboard
59.0%
pass@1 · highest published configuration
mini-swe-agent · max reasoning
pass@4
79.3%
Average cost / attempt
$13.22
Benchmark runs
4

All published configurations

Harness / reasoningpass@1pass@4Cost / attempt
mini-swe-agent · max59.0%79.3%$13.22
mini-swe-agent · xhigh54.4%80.5%$8.01
mini-swe-agent · high51.8%77.9%$4.28
mini-swe-agent · medium48.7%76.1%$3.44
mini-swe-agent · low40.8%68.1%$2.29

pass@1 is the success rate across scored attempts; pass@4 is the share of tasks solved in at least one of four attempts. Costs are reported by DeepSWE for each configuration. These results are not directly comparable to evaluations using different harnesses. Official source JSON.

Other published benchmarks

EvaluationReported score and source
ARC-AGI-2
72.1%
Third-party evaluator · claude-opus-4-8_high
Reasoning effort: high
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
62.2%
Third-party evaluator · claude-opus-4-8_low
Reasoning effort: low
External evaluation via Epoch AI
Original-source terms apply
ARC-AGI-2
71.7%
Third-party evaluator · claude-opus-4-8_medium
Reasoning effort: medium
External evaluation via Epoch AI
Original-source terms apply
FrontierMath-2025-02-28-Private
47.2%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
FrontierMath-Tier-4-2025-07-01-Private
31.3%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
FrontierMath-Tier-4-v2-Private
56.1%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
FrontierMath-Tiers-1-3-v2-Private
80.0%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
GPQA diamond
88.4%
Epoch AI · claude-opus-4-8_low
Reasoning effort: low
Epoch AI evaluation
GPQA diamond
91.0%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
GPQA diamond
85.4%
Epoch AI · claude-opus-4-8_none
Reasoning effort: none
Epoch AI evaluation
OSWorld 2.0
18.5%
https://osworld-v2.xlang.ai/ · claude-opus-4-8_max
Reasoning effort: max
https://osworld-v2.xlang.ai/
Original-source terms apply
OTIS Mock AIME 2024-2025
97.8%
Epoch AI · claude-opus-4-8_low
Reasoning effort: low
Epoch AI evaluation
OTIS Mock AIME 2024-2025
98.3%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation
OTIS Mock AIME 2024-2025
84.4%
Epoch AI · claude-opus-4-8_none
Reasoning effort: none
Epoch AI evaluation
SimpleQA Verified
53.0%
Epoch AI · claude-opus-4-8_max
Reasoning effort: max
Epoch AI evaluation

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Specifications

Developer
Anthropic
Model family
Claude
Type
Language model
Predecessor
—
Successor
—
Released
May 28, 2026
Model ID
anthropic/claude-opus-4.8
Openness
Closed
License
—
Input modalities
text, image, file
Output modalities
text
Modality
text+image+file->text
Tokenizer
Claude
Max output tokens
128K
Knowledge cutoff
2026-01
Hugging Face
—
Source freshness
Metadata synced Oct 8, 2026, 10:17 PM UTC
Additional source evidence
Source model
anthropic/claude-opus-4-8
Source family
claude-opus
Last updated
2026-05-28
Knowledge cutoff
2026-01
Model context
1M
Model output limit
128K
Open weights
No
File attachments
Yes
Reasoning
Yes
Tool calling
Yes
Temperature control
No

Published benchmarks

BenchmarkScoreEvaluation detailsEvidence
SWE-Bench Pro69.2 · resolve rate2026-05-28Source
Terminal-Bench74.6 · success rateTerminus-2 · 2.1 · 2026-05-28Source
SWE-Bench Verified88.6 · resolved—Source
Humanity's Last Exam49.8 · accuracyno tools · 2026-06-09Source
Humanity's Last Exam57.9 · accuracywith tools · 2026-06-09Source
OSWorld-Verified83.4 · success rate2026-06-09Source
FrontierCode13.4 · pass rateDiamond · high effort · 2026-06-09Source
Complete source record (JSON)
{
  "id": "anthropic/claude-opus-4-8",
  "name": "Claude Opus 4.8",
  "description": "Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents",
  "family": "claude-opus",
  "last_updated": "2026-05-28",
  "knowledge": "2026-01",
  "attachment": true,
  "reasoning": true,
  "tool_call": true,
  "temperature": false,
  "modalities": {
    "input": [
      "text",
      "image",
      "pdf"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 1000000,
    "output": 128000
  },
  "benchmarks": [
    {
      "name": "SWE-Bench Pro",
      "score": 69.2,
      "source": "https://www.anthropic.com/news/claude-opus-4-8",
      "metric": "resolve rate",
      "date": "2026-05-28"
    },
    {
      "name": "Terminal-Bench",
      "score": 74.6,
      "source": "https://www.anthropic.com/news/claude-opus-4-8",
      "metric": "success rate",
      "harness": "Terminus-2",
      "version": "2.1",
      "date": "2026-05-28"
    },
    {
      "name": "SWE-Bench Verified",
      "score": 88.6,
      "source": "https://benchlm.ai/benchmarks/sweVerified",
      "metric": "resolved"
    },
    {
      "name": "Humanity's Last Exam",
      "score": 49.8,
      "source": "https://www.anthropic.com/news/claude-fable-5-mythos-5",
      "metric": "accuracy",
      "variant": "no tools",
      "date": "2026-06-09"
    },
    {
      "name": "Humanity's Last Exam",
      "score": 57.9,
      "source": "https://www.anthropic.com/news/claude-fable-5-mythos-5",
      "metric": "accuracy",
      "variant": "with tools",
      "date": "2026-06-09"
    },
    {
      "name": "OSWorld-Verified",
      "score": 83.4,
      "source": "https://www.anthropic.com/news/claude-fable-5-mythos-5",
      "metric": "success rate",
      "date": "2026-06-09"
    },
    {
      "name": "FrontierCode",
      "score": 13.4,
      "source": "https://www.anthropic.com/news/claude-fable-5-mythos-5",
      "metric": "pass rate",
      "variant": "high effort",
      "dataset": "Diamond",
      "date": "2026-06-09"
    }
  ],
  "release_date": "2026-05-28",
  "open_weights": false
}

Infrastructure

Provider availability

All providers
ProviderEndpointContextInput /M tokensOutput /M tokensCached input /M tokensCache write /M tokensUptime 1dData handling
Amazon Bedrockamazon-bedrock1M$5.00$25.00$0.50$6.25100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrockBatch pricing1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Anthropicanthropic1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

AnthropicanthropicBatch pricing1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/global1M$5.00$25.00$0.50$6.25—
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/globalBatch pricing1M$5.00$25.00$0.50$6.25—
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Claude Platform on AWSclaude-on-aws1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Claude Platform on AWSclaude-on-awsBatch pricing1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/global1M$5.00$25.00$0.50$6.25100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/globalBatch pricing1M$5.00$25.00$0.50$6.25100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/us1M$5.50$27.50$0.55$6.88100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/eu-west-11M$5.50$27.50$0.55$6.88100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/usBatch pricing1M$5.50$27.50$0.55$6.88100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Amazon Bedrockamazon-bedrock/eu-west-1Batch pricing1M$5.50$27.50$0.55$6.88100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/us1M$5.50$27.50$0.55$6.88—
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Azureazure/usBatch pricing1M$5.50$27.50$0.55$6.88—
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/us1M$5.50$27.50$0.55$6.88—
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/europe1M$5.50$27.50$0.55$6.88100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/usBatch pricing1M$5.50$27.50$0.55$6.88—
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/europeBatch pricing1M$5.50$27.50$0.55$6.88100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Anthropicanthropic/fast1M$10.00$50.00$1.00$12.50100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Anthropicanthropic/fastBatch pricing1M$10.00$50.00$1.00$12.50100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Provider data:OpenRouter

Text provenance

Transitioning

Claude text watermarking

Anthropic says Claude models launched before August 2, 2026 have a transition period and are being moved to the same watermarking system over time.

Official explanation