Provider
Anthropic
Base model family
Model intelligence profile
Claude Opus 4.8 benchmark results and model details
This family profile brings together 5 published benchmark records for Anthropic Claude Opus 4.8. Every result keeps its original benchmark, configuration, scale, and source; unrelated scores are never averaged.
Published benchmark evidence available · Business Skills V3 not yet authorized
Release date
2026-05-28
Exact-family source only
Published benchmarks
5
Native scales kept separate
Artificial Analysis
Reference available
As of 2026-07-16
Benchmark-level evidence
Published Claude Opus 4.8 benchmark results
These are individual benchmarks, not collection rollups. Scores remain on their original scales, and agent or harness results stay labelled as system configurations.
Spring Prompt benchmark
BulletBench
304
60-second bullet Elo
Chess decision quality under a shared 60-second game clock and a 10+1 per-move clock.
Ladder Elo: anchored to a Stockfish skill ladder (random mover = 400). Internally consistent ordering, not FIDE-calibrated.
1 published configuration
Claude Opus 4.8 (low)
304
- Bullet 95% interval
- 168–416
- 10+1 Elo
- 0
- Bullet games
- 96
- Bullet median move time
- 1.52s
- Average bullet game cost
- $0.0896
Spring Prompt benchmark
ROASBench
3 configurations
Average ROASBench score
A 12-month performance-marketing simulation scored on business outcomes, planning, behavior, and persona fit.
3 published configurations
Anthropic: Claude Opus 4.8
9.54
- Average contribution profit
- $-312,597
- Average ROAS
- 0.87×
- Completed runs
- 3
- Score variability
- ±0.96
Anthropic: Claude Opus 4.8 · High
10.41
- Average contribution profit
- $-352,542
- Average ROAS
- 0.84×
- Completed runs
- 3
- Score variability
- ±2.28
Anthropic: Claude Opus 4.8 · Low
9.98
- Average contribution profit
- $-299,490
- Average ROAS
- 0.89×
- Completed runs
- 3
- Score variability
- ±2.08
Source: Spring Prompt
Recorded 25 Jul 2026
Published ROASBench cache · 12-month simulation
Source record →
Spring Prompt benchmark
World Cup prediction benchmark
296
Tournament prediction points
Fixture-by-fixture football predictions frozen before the tournament and graded on outcomes, scorelines, goals, penalties, and cards.
32 fixtures graded from the frozen prediction artifact.
1 published configuration
Claude Opus 4.8
296
- Graded matches
- 32
- Exact-score points
- 15
- Result points
- 36
- Goal points
- 66
source-native benchmark
Coding Agent Configurations
2 configurations
Coding Agent Index v1.2 point score
Artificial Analysis Coding Agent Index v1.2 point scores and task-specific operational measurements for exact agent and model configurations.
These are exact agent-plus-model configurations, not bare-model scores.
2 published configurations
Claude Code - Opus 4.8 (max)
54.9
- Mean cost per task
- $7.70
- Mean wall time
- 23.1 min
Claude Code - Opus 4.8 (medium)
48.9
- Mean cost per task
- $3.26
- Mean wall time
- 12.4 min
source-native benchmark
Repository Issue Resolution
83.8%
OpenHands SWE-Bench resolved
Direct OpenHands SWE-Bench outcomes for resolving real repository issues with a pinned OpenHands agent and language-model configuration.
Scores are official OpenHands SWE-Bench resolved percentages for exact OpenHands and model configurations, not bare-model scores or SpringPrompt predictions. Twenty-nine exact task sidecars support marginal 95% bootstrap score intervals; five rows explicitly have no interval. No full-cohort rank confidence is claimed.
1 published configuration
OpenHands v1.24.0 + claude-opus-4-8
83.8%
- 95% task-bootstrap interval
- 80.6–87.0%
- Task evidence
- exact-headline-reconstruction
Source: OpenHands Index
Recorded 30 Jun 2026
OpenHands · SWE-Bench 2026.06.30-3015ac6
Source record →
Independent operational reference
Artificial Analysis details
Representative configuration: Claude Opus 4.8 (Adaptive Reasoning, Max Effort). Lifecycle: active; released 2026-05-28.
Input price
$5.00 / 1M tokens
Output price
$25.00 / 1M tokens
Median output speed
54.8 tok/s
Operational reference only. Price and throughput are not model-quality scores and never affect Spring Prompt comparisons.
Early-user signal
Initial community opinions
A paraphrased editorial synthesis of 4 Reddit discussions created in the 14 days after launch (28 May 2026 to 11 Jun 2026). Anecdotal context only — it never affects scores or rankings.
The search-indexed launch-window sample was sharply divided on Claude Opus 4.8. Some users reported more reliable tool use and longer stable engineering runs, while others encountered verbosity, slow inference, language errors, repeated actions, and rapid quota use. Experiences appeared to differ across languages, interfaces, effort settings, and task types.
Independent early-test signal
Early technical field tests
An editorial paraphrase of 2 launch-window field tests from 2 independent authors on X, including 2 reports with a described method or inspectable artifact. The window runs from 28 May 2026 up to 11 Jun 2026, and exact model identity was editor reviewed.
Early Opus 4.8 evidence combined an autonomous, long-horizon RPG build with two agent-benchmark results that moved in different directions. The artifact demonstrated broad planning, document creation, play-testing, and deployment, while the benchmark report found regressions on simulated business and planning tasks. The safest interpretation is high artifact-building ability with uneven agent performance by task and reasoning setting.
These reports are selectively surfaced and are not a representative sample. They never affect benchmark scores, rankings, winners, or comparison outcomes.
Business Skills V3 · proposed
How Spring Prompt plans to test Claude Opus 4.8
The setup below is proposed and may change until preflight and execution approval are complete. Existing benchmark evidence above does not authorize or stand in for a Business Skills V3 result.
- Proposed reasoning
- High reasoning
- Provider revision
- Exact provider revision will be resolved and frozen only after preflight and execution approval
- Tools and service
- No tools · Anthropic standard routing
- Planning configuration
- claude-opus-4.8-high
- Generation controls
- Provider-managed reasoning; no temperature override proposed
Future first-party coverage
15 proposed Business Skills V3 task areas
These links describe evaluation contracts, not published Claude Opus 4.8 results.
Professional Knowledge Work & StrategyBusiness Workflow AutomationCustomer Support & Service ResolutionSales & MarketingLegal Document WorkFinance, Accounting & SpreadsheetsData Analysis & SQLResearch & Evidence SynthesisBusiness Writing & EmailDocument Comprehension & ExtractionSummarization & Grounded KnowledgeSoftware Engineering & CodingFrontend, UI & Web AppsTool Calling & Structured OutputPresentations, Charts & Visual Communication
Model comparisons
Compare Claude Opus 4.8 side by side
Editorially reviewed comparisons appear first. Every page matches only benchmark records with the same reviewed protocol key and does not manufacture an overall winner.
Publication safeguards
What must pass before a V3 result appears
- 1First-party task-local comparisons and eligible external evidence must both be present.
- 2Model and provider configuration identity must match the reviewed release exactly.
- 3Coverage, reliability, judge diagnostics, and sealed stability checks must pass.
- 4Uncertainty and missing evidence remain visible when results are published.
Stable family URL
Evidence can grow without changing the page
New reviewed benchmark snapshots, operational facts, community themes, and early field-test syntheses can be added here while the canonical model-family identity remains fixed.
More from Anthropic