Claude Opus 5 benchmarks
Claude Opus 5 is a proprietary model from Anthropic, released 24 Jul 2026. It ranks 4th of 51 on our overall leaderboard and 3rd of 42 for coding. At A$14.04 per million tokens (blended), it costs 3.9× the median model we track. It writes about 51 tokens a second, and answers start after 34 s on average, thinking included. It can run with requests kept in Australia on AWS Bedrock in Sydney and Melbourne.
- CONTEXT WINDOW
- 1,000,000 tokens
- MAX OUTPUT
- 128,000 tokens
- INPUT
- text, image
- WEIGHTS
- Proprietary
- RELEASED
- 24 Jul 2026
- OUTPUT SPEED
- 51 tok/s
- ANSWER STARTS AFTER
- 34 s (thinking included)
Claude Opus 5 across our rankings.
Benchmark results.
Independent test scores are for the adaptive reasoning, max effort setting. “Among tracked” ranks Claude Opus 5 against the 51 models on this site; LMArena ranks run across its full leaderboard.
| BENCHMARK | RESULT | AMONG TRACKED | SOURCE DETAIL |
|---|---|---|---|
| INDEPENDENT TESTS · ARTIFICIAL ANALYSIS | |||
| AA Intelligence Index | 50.8 | #4 of 51 | Adaptive reasoning, max effort |
| AA Coding Index | 78.0 | #2 of 42 | Adaptive reasoning, max effort |
| GPQA Diamond | 93.2% | #11 of 44 | Adaptive reasoning, max effort |
| Humanity’s Last Exam | 54.9% | #4 of 51 | Adaptive reasoning, max effort |
| AA-LCR | 79.3% | #34 of 51 | Adaptive reasoning, max effort |
| Terminal-Bench 2.1 | 89.1% | #2 of 42 | Adaptive reasoning, max effort |
| τ²-Bench (banking) | 42.1% | #12 of 41 | Adaptive reasoning, max effort |
| SciCode | 56.4% | #19 of 46 | Adaptive reasoning, max effort |
| BLIND HUMAN VOTES · LMARENA | |||
| LMArena Text42,617 votes · ±4 | 1493 | #8 of 42 | #10 of 402 on LMArena · 13 Sep 2026 |
| LMArena WebDev14,251 votes · ±7 | 1691 | #3 of 45 | #3 of 129 on LMArena · 22 Sep 2026 |
| LMArena Vision11,599 votes · ±8 | 1289 | #8 of 33 | #10 of 152 on LMArena · 13 Sep 2026 |
| LMArena Document8,794 votes · ±7 | 1490 | #5 of 26 | #7 of 44 on LMArena · 13 Sep 2026 |
| LMArena Agent24,794 sessions | #3 | #3 of 37 | #3 of 46 on LMArena · 15 Sep 2026 |
What Claude Opus 5 costs in Australian dollars.
List prices converted at the snapshot’s RBA rate, excluding GST. Reasoning models are billed for their thinking as output tokens, so heavier settings cost more than the job estimates show.
| PER MILLION TOKENS | AUD | USD LIST |
|---|---|---|
| Input tokens | A$7.02 | US$5.00 |
| Output tokens | A$35.10 | US$25.00 |
| Blended (3 in : 1 out) | A$14.04 | US$10.00 |
| JOB | COST (AUD) | MEDIAN MODEL |
|---|---|---|
| Customer support replyPER 1,000 REPLIES | A$24.57 | ≈ A$6.58 |
| Summarise a 30-page documentPER 100 DOCUMENTS | A$16.85 | ≈ A$4.74 |
| Agentic coding taskPER 10 TASKS | A$13.34 | ≈ A$3.72 |
| SETTING | INTELLIGENCE | A$ / 1M | SPEED |
|---|---|---|---|
| Adaptive reasoning, max effort | 50.8 | A$14.04 | 51 tok/s |
| Adaptive reasoning, xhigh effort | 49.7 | A$14.04 | 50 tok/s |
| Adaptive reasoning, high effort | 48.1 | A$14.04 | 51 tok/s |
| Adaptive reasoning, medium effort | 44.8 | A$14.04 | 51 tok/s |
| Adaptive reasoning, low effort | 39.4 | A$14.04 | 56 tok/s |
Running Claude Opus 5 in Australia.
It can run with requests kept in Australia on AWS Bedrock in Sydney and Melbourne. How the platforms compare →
| PLATFORM | AVAILABILITY | DETAIL |
|---|---|---|
| AWS Bedrock · Sydney | AU ROUTING | |
| AWS Bedrock · Melbourne | IN REGION | In-region through the bedrock-mantle endpoint (Anthropic Messages API); the standard runtime endpoint uses the au. profile. Provider docs, checked 23 Sep 2026 |
| Azure · Australia East | NOT IN AU | |
| Google Vertex AI · Sydney | GLOBAL |
Compare it with.
Claude Opus 5, answered.
How much does Claude Opus 5 cost in Australian dollars?
Claude Opus 5 lists at US$5.00 per million input tokens and US$25.00 per million output tokens — A$7.02 and A$35.10 at A$1 = US$0.7123 (RBA, 22 Sep 2026), excluding GST. A typical customer support reply works out at about A$24.57 per 1,000 replies, before any reasoning tokens.
Can I use Claude Opus 5 in Australia with data kept onshore?
Claude Opus 5 can run with requests kept in Australia on AWS Bedrock in Sydney and Melbourne. Availability comes from the cloud providers’ own documentation; check your provider’s current terms before relying on it for data-residency obligations.
Is Claude Opus 5 good for coding?
It ranks 3rd of 42 on our coding ranking, with an Artificial Analysis Coding Index of 78.0 and an LMArena WebDev rating of 1691 (3rd of 129).
How good is Claude Opus 5 at agent and automation work?
It ranks 3rd of 41 on our agents ranking, completing 42.1% of τ²-Bench banking customer-service tasks and 89.1% of Terminal-Bench 2.1 tasks.
How fast is Claude Opus 5?
Artificial Analysis measures Claude Opus 5 at about 51 tok/s of output, with the answer starting after 34 s on average once thinking time is included (adaptive reasoning, max effort setting).
What is Claude Opus 5’s context window?
Anthropic documents a context window of 1,000,000 tokens (1M), with up to 128,000 output tokens. On AA-LCR, which tests reasoning across ~100,000-token document sets, it scores 79.3%.
Ratings: LMArena leaderboard dataset (CC BY 4.0), rescaled for the lens scores · leaderboards to 22 Sep 2026
Evaluations, prices and speed: Artificial Analysis (artificialanalysis.ai) · fetched 23 Sep 2026
Exchange rate: Reserve Bank of Australia, table F11.1 · A$1 = US$0.7123 on 22 Sep 2026 · prices exclude GST
Snapshot 23 Sep 2026 · updated weekly · How the rankings work →