Across 17 comparable benchmarks, Claude Fable 5.2 has better reported results on 12 and worse on 5. 1 further shared result is listed below, each with the reason it is not counted.
Moderate evidence17 of 18 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | Gemini 4 Neon | Claude Fable 5.2 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | $3.00 | $8.00 |
| Output / 1M | $15.00 | $40.00 |
| Blended (4:1) | $3.35 | $8.78 |
| Status | Available | Available |
| Released | November 18, 2026 | November 9, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates β set your own mix.
Mean of 10 shared benchmarks grouped by capability on a 0β100 scale. How averages are built
Claude Fable 5.2 ahead on 4 of 4 counted
Claude Fable 5.2 ahead on 3 of 3 counted
Gemini 4 Neon ahead on 1 of 1 counted
Split 1β1 of 2 counted
Gemini 4 Neon ahead on 3 of 4 counted
No shared comparable results for Abstract reasoning, Mathematics, Research & search, Long context, Cybersecurity.
Published scores, with matching modes where available. Gaps show Gemini 4 Neon relative to Claude Fable 5.2. Reading the results
| Benchmark | Gemini 4 Neon | Claude Fable 5.2 | Gap |
|---|---|---|---|
| FrontierCode 2.0Repository Engineering Β· Weighted rubric score | 46.2Thinking High Β· tools + web | 49.1β Extra-High Β· tools | 0.94Γ#3 vs #1 of 32 |
| Terminal-Bench 4.0Agentic Development Β· Resolution rate (%) | 63.6Extra-High Β· tools | 67.6β Max Β· tools | β4.0 pts#7 vs #3 of 90 |
| Terminal-Bench 4.1Agentic Development Β· Resolution rate (%) | 65.2Extra-High Β· tools | 65.8β Max Β· tools | β0.6 pts#3 vs #1 of 48 |
| Vibe Code Bench v1.1Code Generation & Editing Β· Pass rate (%) | 93.7Extra-High Β· tools | 93.8β Thinking High Β· tools | β0.1 pts#2 vs #1 of 119 |
| Vending-Bench 2Logical Planning Β· Money Balance (USD)Not counted Β· Speed and cost telemetry is reported, not scored | 16843Unspecified Reasoning Β· tools | 9628Unspecified Reasoning Β· tools | 1.75Γ#1 vs #13 of 64 |
| MysteryMechanismScientific Reasoning Β· Accuracy (%) | 45.0Extra-High Β· tools | 51.8β Thinking High Β· tools | β6.8 pts#9 vs #3 of 32 |
| Terminal-Bench-Science 0.1Scientific Computing Β· Task success rate (%) | 70.0Extra-High Β· tools | 74.6β Max Β· tools | β4.6 pts#4 vs #3 of 32 |
| BioMysteryBenchBiology & Genomics Β· Accuracy (%) | 80.0Extra-High Β· tools | 83.7β Thinking High Β· tools | β3.7 pts#6 vs #3 of 28 |
| AutomationBench-AATool Orchestration Β· Guardrail-safe objective completion rate | 79.3Extra-High Β· tools | 76.4β Max Β· tools | +2.9 pts#1 vs #5 of 15 |
| ChartographyDocuments & Charts Β· Score | 73.5Extra-High Β· no tools | 91.6β Max Β· tools | 0.80Γ#15 vs #1 of 24 |
| GDP.pdfDocuments & Charts Β· Task pass rate (%) | 33.4Thinking High Β· no tools | 26.8β Thinking Low Β· no tools | +6.6 pts#3 vs #14 of 93 |
| AA-BriefcaseOffice & Business Β· AA-Briefcase Elo | 1700Extra-High Β· tools | 1785β Max Β· tools | 0.95Γ#5 vs #3 of 17 |
| AutomationBenchOffice & Business Β· Pass Rate | 56.3Extra-High Β· toolsΒ· zapier-private-heldout | 51.5β Max Β· toolsΒ· zapier-private-heldout | +4.8 pts#3 vs #10 of 64 |
| Legal Research BenchLegal Β· Score | 56.3Extra-High Β· tools + internet | 46.2β Thinking High Β· tools + web | 1.22Γ#1 vs #19 of 91 |
| Tax Agent BenchFinance Β· Accuracy (%, rubric Γ citation quality) | 54.9Extra-High Β· tools + internet | 32.1β Thinking High Β· tools + web | +22.8 pts#22 vs #36 of 74 |
| HLE-DiamondKnowledge Exams Β· Accuracy (%) | 71.3Extra-High Β· no tools | 80.2β Max Β· no tools | β8.9 pts#3 vs #1 of 11 |
| AA Intelligence Index v5.0.1Capability Indices Β· Index (0-100) | 61.4Extra-High Β· tools | 64.6β Max Β· task tools | 0.95Γ#3 vs #1 of 61 |
| Vals Index v2.2Capability Indices Β· GDP-weighted accuracy (%) | 70.9Extra-High Β· tools | 71.3β Thinking High Β· tools | β0.4 pts#2 vs #1 of 34 |
Repository Engineering Β· Weighted rubric score
Agentic Development Β· Resolution rate (%)
Agentic Development Β· Resolution rate (%)
Code Generation & Editing Β· Pass rate (%)
Logical Planning Β· Money Balance (USD)
Not counted Β· Speed and cost telemetry is reported, not scored
Scientific Reasoning Β· Accuracy (%)
Scientific Computing Β· Task success rate (%)
Biology & Genomics Β· Accuracy (%)
Tool Orchestration Β· Guardrail-safe objective completion rate
Documents & Charts Β· Score
Documents & Charts Β· Task pass rate (%)
Office & Business Β· AA-Briefcase Elo
Office & Business Β· Pass Rate
Legal Β· Score
Finance Β· Accuracy (%, rubric Γ citation quality)
Knowledge Exams Β· Accuracy (%)
Capability Indices Β· Index (0-100)
Capability Indices Β· GDP-weighted accuracy (%)
A dash means no result is available.
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix Β· 5% steps
Blended price = effective input rate at 90% cache hits Γ 80% + output rate Γ 20%
Gemini 4 Neon costs62%lessthan Claude Fable 5.2 per million tokens β 2.6Γ cheaper.Using standard rates at this token mix.
Gemini 4 Neon: Standard pricing. Unified 1M context pricing; thinking tokens included in output; paid tier only (no free tier).
Claude Fable 5.2: Standard pricing. Global first-party pricing; US-only inference adds 10%.
| Price / 1M tokens | Gemini 4 Neon | Claude Fable 5.2 | Ratio |
|---|---|---|---|
| Input | $3.00 | $8.00 | 0.38Γ |
| Output | $15.00 | $40.00 | 0.38Γ |
| Cached input | $0.15 | $0.20 | 0.75Γ |
| Cache write | β | β | β |
| Blended | $3.35 | $8.78 | 0.38Γ |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
Gemini 4 Neon relative to Claude Fable 5.2
Summary: 1 of 5 shared benchmarks have matching recorded settings and sources. The average is a geometric mean of 1 positive-cost ratios, one equal weight per benchmark. This benchmark set can include related tests and indexes; it is not a workload cost estimate.
Gemini 4 Neon 2.4Γ cheaper
Settings or sources differ
Settings or sources differ
Settings or sources differ
Settings or sources differ
Gemini 4 Neon 2.4Γ cheaper
Settings or sources differ
Settings or sources differ
Settings or sources differ
Settings or sources differ
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 4 Neon
Claude Fable 5.2
Settings or sources differ or are not recorded; interpret the comparison with care.
Gemini 4 Neon
Claude Fable 5.2
Settings or sources differ or are not recorded; interpret the comparison with care.
Gemini 4 Neon
Claude Fable 5.2
Settings or sources differ or are not recorded; interpret the comparison with care.
Gemini 4 Neon
Claude Fable 5.2
Settings or sources differ or are not recorded; interpret the comparison with care.
Gemini 4 Neon
Claude Fable 5.2
| Benchmark | Gemini 4 Neon | Claude Fable 5.2 |
|---|---|---|
| AA Coding Agent Index v2.0 | Not reported | |
| AA-Briefcase | Not reported | 2 more settings |
| Agent Arena | Not reported | $8.464 / task Median Β· Net improvement index (percentage points vs. average model): 30.5 Max (With Tools + Internet) |
| ARC-AGI-1 | Not reported | 2 more settings |
| ARC-AGI-2 | Not reported | 2 more settings |
| AutomationBench | Not reported | $1.69 / task Average Β· Pass Rate: 44.3 Thinking High (With Tools) Β· zapier-private-heldout $1.30 / task Average Β· Pass Rate: 39.7 Thinking Low (With Tools) Β· zapier-private-heldout 2 more settings$1.35 / task Average Β· Pass Rate: 40.2 Thinking Medium (With Tools) Β· zapier-private-heldout $2.29 / task Average Β· Pass Rate: 48.0 Extra-High (With Tools) Β· zapier-private-heldout |
| BenchCAD 2.1 Core | Not reported | $50.77 / task Average Β· Core mean geometry/circuit score (0β100): 83.5 Max (With Tools) $9.993 / task Average Β· Core mean geometry/circuit score (0β100): 61.0 Thinking High (With Tools) $1.888 / task Average Β· Core mean geometry/circuit score (0β100): 26.9 Thinking Low (With Tools) 2 more settings$2.66 / task Average Β· Core mean geometry/circuit score (0β100): 32.4 Thinking Medium (With Tools) $23.23 / task Average Β· Core mean geometry/circuit score (0β100): 74.1 Extra-High (With Tools) |
| BioMysteryBench | Not reported | |
| Chartography | Not reported | 3 more settings |
| Code Migration | Not reported | |
| CursorBench 4.0 | Not reported | $32.79 / task Average Β· Task score: 60.0 Max (With Tools) Β· cursorbench-4-0-cursor-agent $9.70 / task Average Β· Task score: 56.9 Thinking High (With Tools) Β· cursorbench-4-0-cursor-agent $3.94 / task Average Β· Task score: 53.2 Thinking Low (With Tools) Β· cursorbench-4-0-cursor-agent 2 more settings$4.66 / task Average Β· Task score: 54.8 Thinking Medium (With Tools) Β· cursorbench-4-0-cursor-agent $15.48 / task Average Β· Task score: 56.3 Extra-High (With Tools) Β· cursorbench-4-0-cursor-agent |
| EMB (Excel Modeling) | Not reported | |
| FrontierCode 2.0 | Not reported | 7 more settings$6.721 / task Average Β· Weighted rubric score: 45.3 Thinking High (With Tools + Internet) $2.098 / task Average Β· Weighted rubric score: 34.2 Thinking Low (With Tools + Internet) $2.482 / task Average Β· Weighted rubric score: 37.6 Thinking Medium (With Tools + Internet) |
| GDP.pdf | Not reported | 2 more settings |
| Harvey's Legal Agent Benchmark | Not reported | |
| HLE | Not reported | 2 more settings |
| HLE-Diamond | Not reported | 2 more settings |
| MysteryMechanism | Not reported | |
| ProofBench v1.1 (Lean 4) | Not reported | |
| Terminal-Bench 4.0 | Not reported | 1 more settings |
| Terminal-Bench 4.1 | Not reported | |
| Terminal-Bench-Science 0.1 | Not reported | 1 more settings |
| WANDR | Not reported | 2 more settings$15.71 / task Average Β· Soft F1 / relaxed score (%): 73.6 Thinking Medium (With Tools) |
| Attribute | Gemini 4 Neon | Claude Fable 5.2 |
|---|---|---|
| Availability | Available | Available |
| Provider | Google Deep Mind | Anthropic |
| Model type | Multimodal model | Reasoning model |
| First release | November 18, 2026 | November 9, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | 1M tokens | 128K tokens |
| Architecture | Undisclosed | Undisclosed |
| Documented capabilities | Reasoning, Coding, Tool use, Computer use, Long context, Function calling, Structured output, Web search | Reasoning, Coding, Tool use, Computer use, Web search, Long context |
| Knowledge cutoff | 2026-03 | 2026-06 |
| Input modalities | Text, Image, Audio, Video | Text, Image |
| Output modalities | Text | Text |
| Weights | Closed / API only | Closed / API only |
| Commercial use | Proprietary | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Gemini 4 Neon or Claude Fable 5.2? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.