Claude Opus 4.8

Name: Claude Opus 4.8 — independent benchmark scores
Creator: Epoch AI
License: https://creativecommons.org/licenses/by/4.0/

Azure AIChat200K context· released 2026-05-28Floating alias

azure_ai/claude-opus-4-8

Estimate cost Compare

Prices as of 2026-07-14 · rev 179affb· unknown is shown as unknown, never $0.00

Strongest atReasoning · Mathematics · Code & agentictop-quartile among comparable models scored on these domains

Value position90th pctile GPQA11 cheaper models match or beat it on GPQA — see the value chart

Data confidence5/5 exactbenchmarks measured on this exact model · prices 2026-07-14

Cost / request$0.01751K in · 500 out

Input / M$5.00

Output / M$25.00

Cache read / M$0.50

Context200K200,000 tokens

Max output128K128,000 tokens

Capability fingerprint

independent benchmarks, grouped by domain · higher is better · not our measurement

Reasoning

GPQA Diamondgraduate-level science reasoning

88%90th pctile

exact modeloptimized runEpoch evaluations · Claude Opus 4.8

SimpleBencheveryday reasoning (unsaturated)

58%85th pctile

exact modelhttps://lmcouncil.ai/benchmarks · Claude Opus 4.8

Mathematics

AIME (OTIS mock)olympiad math

98%96th pctile

exact modeloptimized runEpoch evaluations · Claude Opus 4.8

Code & agentic

WeirdMLnovel out-of-distribution ML coding

83%97th pctile

exact modelhttps://htihle.github.io/weirdml.html · Claude Opus 4.8

Factuality

SimpleQAfactual accuracy (higher = fewer hallucinations)

40%52nd pctile

exact modelEpoch evaluations · Claude Opus 4.8

peer median · percentile ranks this model among distinct comparable models (deduped — one entry per model, not per provider) scored on that benchmark · exact model / family proxy per row.

Source: Epoch AI — “AI Benchmarking Hub” (epoch.ai), CC-BY (Epoch AI). Some rows carry an upstream leaderboard (shown per row); Aider & Terminal-Bench are additionally Apache-2.0. Each benchmark measures a different thing on a different scale — rows are not comparable across domains, and a score is not a guarantee of quality on your task.

Capabilities Index

Epoch AI's cross-model capability re-fit · one comparable axis · not our measurement

158.0ECI

96th percentile of 150 modelsexact model

The dot is Claude Opus 4.8; the whisker is its 95% CI; the firm tick is the peer median; faint ticks are every other scored model. Unlike a raw benchmark, the ECI axis is comparable across models — but it's a relative re-fit scale, so a model's exact position can shift when Epoch recomputes it.

Model facts · Epoch AI, confident confidence

Developer: Anthropic
First published: 2026-05-28
Access: API access

Source: Epoch AI — Capabilities Index & Notable-Models data (epoch.ai), CC-BY (Epoch AI). The ECI is Epoch's index, not TokenTriage's; per-fact confidence labels are Epoch's own.

Human preference (LMArena)

blind pairwise human votes · Elo · style-controlled · not a correctness score

exact modelmeasured on claude-opus-4-8statistically tied with 12 models overall

Each dot is Claude Opus 4.8's Arena score in that category; the whisker is the 95% CI; the firm tick is the peer median across distinct models; faint ticks are the field. This is human preference — which answer people pick in a blind A/B — not a correctness or capability score, and adjacent ranks routinely overlap within their intervals.

Source: LMArena (LMArena Leaderboard (lmarena.ai) — lmarena-ai/leaderboard-dataset on Hugging Face, CC BY 4.0.). Ratings are the style-controlled estimates (length/formatting confound removed), as of 2026-07-27.

Cost vs quality

where Claude Opus 4.8 sits among GPQA-scored models

Each dot is a model with a GPQA Diamond score, priced at a 1K-in / 500-out request; Claude Opus 4.8 is highlighted. The shaded corner holds models that are both cheaper and higher-scoring; the dashed staircase is the value frontier. Hover any dot for its model.

11 cheaper models match or beat Claude Opus 4.8 on GPQA— see the full table below, switchable by benchmark.

Also worth considering

cheaper models that score about as well — pick the benchmark that matters to you

8 cheaper models score within 3 points of Claude Opus 4.8 on GPQA Diamond.

Model	GPQA Diamond	Cost / req	Savings
DEDeepseek V4 Pro	86%-2	$0.00087	−95%
XAGrok 4.3	85%-3	$0.0025	−86%
XAGrok 4.3 Latest	85%-3	$0.0025	−86%
AAKimi K2.6	88%0	$0.00295	−83%
CLKimi K2.7 Code	86%-2	$0.00295	−83%
CLGlm 5.2	89%+1	$0.0036	−79%
AOGPT 5.6 Luna	89%+1	$0.004	−77%
XAGrok 4.20.0309 Reasoning	86%-2	$0.005	−71%

Deduped to one entry per model (cheapest offering), priced at the reference 1K-in / 500-out request. Filtered only on the benchmark score — a model with an unknown capability is never silently excluded. A lower price is only cheaper if quality holds on your task — that's what routing proves.

Pricing

Input · fresh prompt tokens	$5.00 / M
Output · generated tokens	$25.00 / M
Cache read · cached-input hit	$0.50 / M
Cache write · 5-minute TTL	$6.25 / M
Cache write · 1-hour TTL	$10.00 / M
Reasoning · thinking tokens	unknown

Reasoning tokens: no separate rate is published. Providers typically bill thinking tokens at the output rate ($25.00 / M) — which is what the estimator assumes, so an extended-thinking workload isn't silently undercounted.

Model info

Provider: Azure AI
Catalog key: azure_ai/azure_ai/claude-opus-4-8
Identifier: Floating alias (may re-point over time)
Type: Chat
Max input: 200,000 tokens
Max output: 128,000 tokens
Released: 2026-05-28
Knowledge cutoff: 2026-01
Weights: Proprietary
Price snapshot: 2026-07-14 · rev 179affb

Pricing from TokenTriage's resolved price artifact (TokenTriage resolved price artifact (internal/pricing/data/prices.json)); capability + model metadata is third-party (LiteLLM model_prices_and_context_window.json, models.dev). Unknown means unstated, not absent.

Prompt-cache break-even

reuse the same prefix ≥ 2 times and caching wins

Writing a prefix to cache costs a premium, but each subsequent hit reads at $0.50/M instead of $5.00/M. For this model the crossover is 2 reuses — past that, the same prompt keeps getting cheaper. TokenTriage's semantic cache is built to cross that line for you.

Capabilities

7 documented as supported

Input & output

✓Vision

✓PDF input

?Audio input

?Audio output

API behaviour

?Streaming

✓Structured output

✓Prompt caching

✓Reasoning

Agents & tools

✓Function calling

?Parallel tool calls

?Web search

✓Computer use

✓ supported ·✕ not supported ·? not documented ·⚠ sources disagree. A models.dev tag means the LiteLLM catalog was silent and an independent source supplied the value (lower confidence); ? is never silently turned into a confident ✕.

Estimate a request

disjoint token buckets · input = fresh (uncached) tokens

Preset

Input tokens

Output tokens

Cached input

Cache write

Reasoning

Requests / month

Cost / request—

Projected / month—

How output length drives the bill

at a 1K-token prompt · cost scales linearly with output

Output tokens overtake input spend at ~200 tokens for this model — past that, generation is your bill, whatever the headline input price says. Computed by the same cost engine as the estimate above.

Compare Claude Opus 4.8 with other models →

Route to Claude Opus 4.8

OpenAI-compatible · one base URL

curl https://YOUR-TOKENTRIAGE-HOST/v1/chat/completions \
  -H "Authorization: Bearer $TOKENTRIAGE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "azure_ai/azure_ai/claude-opus-4-8",
    "messages": [{"role": "user", "content": "Hello"}]
  }'from openai import OpenAI

client = OpenAI(
    base_url="https://YOUR-TOKENTRIAGE-HOST/v1",
    api_key="YOUR_TOKENTRIAGE_KEY",
)

resp = client.chat.completions.create(
    model="azure_ai/azure_ai/claude-opus-4-8",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

Point your existing OpenAI client at your TokenTriage host and pass the catalog key as the model. TokenTriage attributes the real cost per request and can route to a cheaper model once it's proven just as good on your traffic.

Data & API

machine-readable · rev 179affb

Pull this model's pricing, capabilities, benchmarks and cross-source provenance programmatically — every field carries the snapshot rev, and unknown is null, never 0. Free to reuse with attribution: benchmarks & Capabilities Index © Epoch AI (CC-BY (Epoch AI)); human-preference Elo © LMArena (CC-BY-4.0); capability & price cross-check via models.dev v2 (MIT) · Portkey (MIT) · TrueFoundry (MIT); open-weights facts via Hugging Face (per-repo license).

azure-ai-azure-ai-claude-opus-4-8.json azure-ai-azure-ai-claude-opus-4-8.csv

A price is a guess until it meets your traffic.

This page tells you what Claude Opus 4.8 charges per token. TokenTriage tells you what it costs on your real requests — and cuts the bill only after proving a cheaper model is just as good.

See the product Back to directory