BlogLLM Benchmark Scores as Published, Side by Side
Claude Fable 5.1 vs Gemini 4 Carbon: Benchmark Scores Side by Side
Every test both Claude Fable 5.1 by Anthropic and Gemini 4 Carbon by Google have a published number for: 15 tests in 10 areas. Each number is as printed, with the page it comes from and the setup it was run under; nothing is ranked or rescored, and two numbers under different setups are two results, not one comparison.
Models Claude Fable 5.1, Gemini 4 CarbonData version 2026103001, as of 2026-10-30
Model
What each developer prints on its own model page; a dash means it prints none.
| Fact | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Released | 2026-09-01 | 2026-10-28 |
| Context window | 1M | - |
| Max output | 128K | - |
| Input | Text, image | - |
| Output | Text | - |
| Weights | Closed | - |
Knowledge Work
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| AutomationBench | 31.4%Anthropic, Claude Fable 5.1 and Mythos 5.1, checked 2026-10-15; printed on 9 more pages | 53.9%Google, Gemini 4 Carbon, checked 2026-10-29 |
| Vending-Bench 2 (final balance in USD) | $5422Google, Gemini 4 Carbon, checked 2026-10-29 | $16484Google, Gemini 4 Carbon, checked 2026-10-29 |
Agentic Coding
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Terminal-Bench 4.0 | 55.8%Anthropic, Claude Fable 5.1 and Mythos 5.1, checked 2026-10-15; printed on 11 more pagesModel setup as reported: Release table; exact effort is unspecified here unless stated. See system card for full protocol. 60.9%Anthropic, Claude Fable 5.1 and Mythos 5.1, checked 2026-10-15Model setup as reported: Second value printed in the Fable 5.1 cell as "60.9% (Mythos 5.1)": filed under the Fable 5.1 heading, the label names Claude Mythos 5.1. | 65.2%Google, Gemini 4 Carbon, checked 2026-10-29 |
| FrontierSWE v2 | 57%Anthropic, System Card: Claude Fable 5.1 & Claude Mythos 5.1, checked 2026-10-25; printed on 5 more pages | 61.2%Google, Gemini 4 Carbon, checked 2026-10-29 |
| SWE-sweep | 2.9%Google, Gemini 4 Carbon, checked 2026-10-29 |
Science and Math
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Terminal-Bench Science 0.1 | 52.6%Anthropic, Claude Fable 5.1 and Mythos 5.1, checked 2026-10-15; printed on 8 more pages | 64.8%Google, Gemini 4 Carbon, checked 2026-10-29 |
| RiemannBench | 65.6%Google, Gemini 4 Argon, checked 2026-10-29; printed on 1 more page | 80%Google, Gemini 4 Carbon, checked 2026-10-29 |
Long Context
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| GraphWalks BFS 256k to 1M (F1) | 65%Google, Gemini 4 Argon, checked 2026-10-29; printed on 1 more page | 96.2%Google, Gemini 4 Carbon, checked 2026-10-29 |
Computer Use
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| OSWorld 2.1 offline subset | 76.8%Google, Gemini 4 Carbon, checked 2026-10-29 |
Multimodal Understanding
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Chartography (no tools) | 42.6%Anthropic, System Card: Claude Fable 5.1 & Claude Mythos 5.1, checked 2026-10-25; printed on 4 more pagesEffort: max; Tools: none | 72.5%Google, Gemini 4 Carbon, checked 2026-10-29 |
Cybersecurity
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| ExploitGym | 30.4%Moonshot AI, Introducing Kimi K3.1 Pro, checked 2026-10-24; printed on 1 more pageEffort: max; Fallback: with fallback | |
| CWE-bench v1 | 58%Google, Gemini 4 Argon, checked 2026-10-29; printed on 2 more pages | 71%Google, Gemini 4 Carbon, checked 2026-10-29; printed on 1 more page |
Safety and Alignment
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Gray Swan IPI (attack success rate after k attempts)attack success rate after k attempts, lower is better | 1%Anthropic, Claude Haiku 5.5 System Card, checked 2026-10-19; printed on 2 more pages | 0%Google, Gemini 4 Carbon: our next frontier model, checked 2026-10-29 |
Engineering and Design
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Blueprint-Bench 2 (normalized score) | 41.9%Google, Gemini 4 Carbon, checked 2026-10-29 | 58.6%Google, Gemini 4 Carbon, checked 2026-10-29 |
Preference and Composite Rankings
| Test | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Artificial Analysis Intelligence Index v5.0 | 49Google, Gemini 4 Carbon, checked 2026-10-29 | 59Google, Gemini 4 Carbon, checked 2026-10-29; printed on 1 more page |
Price
API list prices in US dollars per one million tokens, as printed on the developer's pricing page on the day checked; a dash means the developer lists that price in another unit or not at all. Cost per task is measured by Vals AI.
| Price | Claude Fable 5.1 | Gemini 4 Carbon |
|---|---|---|
| Input | $10Anthropic, Anthropic API pricing, checked 2026-10-15 | |
| Cached input | $0.25Anthropic, Anthropic API pricing, checked 2026-10-15 | - |
| Cache write | - | |
| Cache write, 1 hour | $20Anthropic, Anthropic API pricing, checked 2026-10-15 | - |
| Output | $50Anthropic, Anthropic API pricing, checked 2026-10-15 | |
| Input, batch | $5Anthropic, Anthropic API pricing, checked 2026-10-15 | - |
| Output, batch | $25Anthropic, Anthropic API pricing, checked 2026-10-15 | - |
| Cost per taskVals AI, USD, lower is better | - |
Claude Fable 5.1 has 122 more and Gemini 4 Carbon has 6 more: tests only one of them has a number for, listed on its model page.
More side by side
Every number here is read from the same catalog file as the benchmark table, data version 2026103001. A pair has a page when both models are their developers' flagships and both have a number for at least 10 of the same tests. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.