Compare
Four models, one page
Pick the models you are actually choosing between. Every row is scaled within itself, so a bar means something next to its neighbours and nothing across rows.
| Specification | Llama 4 ScoutMeta | Claude Opus 5Anthropic |
|---|---|---|
| Released | Apr 5, 2025 | Jul 24, 2026 |
| Context window | 1.31M | 1M |
| Max output | 16K | 128K |
| Input / 1M | $0.1 | $5 |
| Output / 1M | $0.3 | $25 |
| Cost per task | $0.01 | $2.25 |
| Arena Elo | 923 | 1,393 |
| Serving providers | 4 | 5 |
| Parameters | 108.6B | Undisclosed |
| Licence | other | Proprietary |
| Capabilities |
|
|
Metrics side by side
Each row is scaled to the largest value in that row — bars compare within a row, never across rows
- Llama 4 Scout
- Claude Opus 5
Intelligence Index
Llama 4 Scout10.3Claude Opus 563.1Coding Index
Llama 4 Scout8.2Claude Opus 578.0Agentic Index
Llama 4 Scout1.1Claude Opus 559.2Output speed
Llama 4 Scout152 t/sClaude Opus 587 t/sContext window
Llama 4 Scout1.31MClaude Opus 51MLatency · lower is better
Llama 4 Scout265msClaude Opus 51.12sBlended price / 1M · lower is better
Llama 4 Scout$0.15Claude Opus 5$10Cost per task · lower is better
Llama 4 Scout$0.01Claude Opus 5$2.25
Rows marked “lower is better” still draw a longer bar for a larger number — read the value, not just the length. Arena Elo is in the specification table above instead: it has no meaningful zero, so a bar would flatten the gaps.
View as table
| Metric | Llama 4 Scout | Claude Opus 5 |
|---|---|---|
| Intelligence Index | 10.3 | 63.1 |
| Coding Index | 8.2 | 78.0 |
| Agentic Index | 1.1 | 59.2 |
| Output speed | 152 t/s | 87 t/s |
| Context window | 1.31M | 1M |
| Latency · lower is better | 265ms | 1.12s |
| Blended price / 1M · lower is better | $0.15 | $10 |
| Cost per task · lower is better | $0.01 | $2.25 |
Evaluation scores
Percentage correct on a common 0–100% scale
- Llama 4 Scout
- Claude Opus 5
GPQA Diamond
Llama 4 Scout58.7%Claude Opus 593.2%Humanity's Last Exam
Llama 4 Scout3.8%Claude Opus 554.9%SciCode
Llama 4 Scout17.0%Claude Opus 555.7%τ²-bench
Llama 4 Scout15.5%Claude Opus 542.1%Terminal-Bench Hard
Llama 4 Scout1.5%Claude Opus 5not measuredLiveCodeBench
Llama 4 Scout29.9%Claude Opus 5not measuredAA-LCR long context
Llama 4 Scout30.3%Claude Opus 575.7%AIME 2025
Llama 4 Scout14.0%Claude Opus 5not measured
A missing bar means that evaluation was not run for that model — it is not a zero.
View as table
| Evaluation | Llama 4 Scout | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | 58.7% | 93.2% |
| Humanity's Last Exam | 3.8% | 54.9% |
| SciCode | 17.0% | 55.7% |
| τ²-bench | 15.5% | 42.1% |
| Terminal-Bench Hard | 1.5% | — |
| LiveCodeBench | 29.9% | — |
| AA-LCR long context | 30.3% | 75.7% |
| AIME 2025 | 14.0% | — |
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).