Across 12 comparable benchmarks, Claude Sonnet 5.5 has better reported results on 12 and worse on 0. 1 further shared result is listed below, each with the reason it is not counted.
Moderate evidence12 of 13 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | Inkling 2 Mini | Claude Sonnet 5.5 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | $0.150 | $2.00 |
| Output / 1M | $0.500 | $10.00 |
| Blended (4:1) | $0.123 | $2.23 |
| Status | Available | Available |
| Released | November 20, 2026 | September 28, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates β set your own mix.
Mean of 6 shared benchmarks grouped by capability on a 0β100 scale. How averages are built
Claude Sonnet 5.5 ahead on 3 of 3 counted
Claude Sonnet 5.5 ahead on 1 of 1 counted
Claude Sonnet 5.5 ahead on 1 of 1 counted
Claude Sonnet 5.5 ahead on 2 of 2 counted
Claude Sonnet 5.5 ahead on 3 of 3 counted
No shared comparable results for Mathematics, Science & health, Research & search, Long context, Cybersecurity.
Published scores, with matching modes where available. Gaps show Inkling 2 Mini relative to Claude Sonnet 5.5. Reading the results
| Benchmark | Inkling 2 Mini | Claude Sonnet 5.5 | Gap |
|---|---|---|---|
| DeepSWERepository Engineering Β· Pass@1 (DeepSWE v1.1) | 48.2Thinking Β· tools | 71.0β Max Β· tools | β22.8 pts#61 vs #23 of 73 |
| FrontierCode 2.0Repository Engineering Β· Weighted rubric score | 15.1Thinking Β· tools | 41.0β Max Β· tools + web | 0.37Γ#26 vs #8 of 36 |
| Vibe Code Bench v1.1Code Generation & Editing Β· Pass rate (%) | 77.9Thinking Β· tools | 92.4β Max Β· tools | β14.5 pts#38 vs #3 of 121 |
| ARC-AGI-1Abstract Generalization Β· Score (%) | 84.5Thinking Β· no tools | 96.5β Extra-High Β· no tools | β12.0 pts#47 vs #13 of 87 |
| ARC-AGI-2Abstract Generalization Β· Score (% solved); cost per task (USD)Not counted Β· Speed and cost telemetry is reported, not scored | 49.3Thinking Β· no tools | 83.9β Extra-High Β· no tools | 0.59Γ#45 vs #19 of 80 |
| AutomationBench-AATool Orchestration Β· Guardrail-safe objective completion rate | 41.8Thinking Β· tools | 71.8β Max Β· tools | β30.0 pts#17 vs #10 of 18 |
| ChartographyDocuments & Charts Β· Score | 72.5Thinking Β· tools | 90.2β Max Β· tools | 0.80Γ#16 vs #3 of 29 |
| GDP.pdfDocuments & Charts Β· Task pass rate (%) | 16.8Thinking Β· no tools | 25.8β Max Β· no tools | β9.0 pts#38 vs #16 of 96 |
| AA-AnalystAgentData Analysis Β· pass^5 (%) | 46.3Thinking Β· tools | 57.5β Max Β· tools + web | β11.2 pts#16 vs #3 of 39 |
| AA-BriefcaseOffice & Business Β· AA-Briefcase Elo | 1263Thinking Β· tools | 1812β Max Β· tools | 0.70Γ#18 vs #1 of 20 |
| GDPval-AA v2Cross-industry Work Β· Elo score | 1319Thinking Β· tools | 1838β Max Β· tools | 0.72Γ#19 vs #4 of 23 |
| AA Intelligence Index v5.0.1Capability Indices Β· Index (0-100) | 20.1Thinking Β· no tools | 54.5β Max Β· task tools | 0.37Γ#44 vs #9 of 62 |
| Vals Index v2.2Capability Indices Β· GDP-weighted accuracy (%) | 46.5Thinking Β· tools | 70.8β Max Β· task tools | β24.3 pts#31 vs #3 of 36 |
Repository Engineering Β· Pass@1 (DeepSWE v1.1)
Repository Engineering Β· Weighted rubric score
Code Generation & Editing Β· Pass rate (%)
Tool Orchestration Β· Guardrail-safe objective completion rate
Documents & Charts Β· Score
Documents & Charts Β· Task pass rate (%)
Data Analysis Β· pass^5 (%)
Office & Business Β· AA-Briefcase Elo
Cross-industry Work Β· Elo score
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%
Inkling 2 Mini costs94%lessthan Claude Sonnet 5.5 per million tokens β 18.2Γ cheaper.Using standard rates at this token mix.
Inkling 2 Mini: Standard pricing. Tinker 64K/256K training (1M sampling); fine-tuning billed separately ($0.75/1M for 64K, $1.50/1M for 256K).
Claude Sonnet 5.5: Standard pricing. Anthropic official pricing (platform.claude.com/docs/en/about-claude/pricing) verified 2026-10-21; full 1M context at standard price
| Price / 1M tokens | Inkling 2 Mini | Claude Sonnet 5.5 | Ratio |
|---|---|---|---|
| Input | $0.15 | $2.00 | 0.07Γ |
| Output | $0.50 | $10.00 | 0.05Γ |
| Cached input | $0.015 | $0.10 | 0.15Γ |
| Cache write | β | β | β |
| Blended | $0.123 | $2.23 | 0.06Γ |
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.
| Benchmark | Inkling 2 Mini | Claude Sonnet 5.5 |
|---|---|---|
| AA Coding Agent Index v1.5 | Not reported | 2 more settings |
| AA Coding Agent Index v2.0 | Not reported | |
| AA Intelligence Index v4.3 | Not reported | 2 more settings |
| AA Intelligence Index v5.0.1 | Not reported | 2 more settings |
| AA-Briefcase | Not reported | 2 more settings |
| AA-Omniscience | Not reported | 2 more settings |
| Agent Arena | Not reported | $3.4989 / task Median Β· Net improvement index (percentage points vs. average model): 16.1 Max (With Tools + Internet) $0.9063 / task Median Β· Net improvement index (percentage points vs. average model): 8 Thinking High (With Tools + Internet) |
| Agent Arena Index | Not reported | $3.4124 / task Median Β· Agent Arena Index rating (index points): 1296 Max (With Tools + Internet) $0.7648 / task Median Β· Agent Arena Index rating (index points): 1196 Thinking High (With Tools + Internet) |
| ARC-AGI-1 | Not reported | 2 more settings |
| ARC-AGI-2 | Not reported | 1 more settings |
| AutomationBench | Not reported | $0.36 / task Average Β· Pass Rate: 29.4 Thinking High (With Tools) Β· zapier-private-heldout $0.28 / task Average Β· Pass Rate: 18.3 Thinking Low (With Tools) Β· zapier-private-heldout 3 more settings$0.31 / task Average Β· Pass Rate: 23.3 Thinking Medium (With Tools) Β· zapier-private-heldout $0.30 / task Average Β· Pass Rate: 18.6 Thinking (With Tools) Β· zapier-private-heldout $0.48 / task Average Β· Pass Rate: 36.8 Extra-High (With Tools) Β· zapier-private-heldout |
| AutomationBench-AA | Not reported | $1.0324 / task Average Β· Guardrail-safe objective completion rate: 71.8 Max (With Tools) $0.3126 / task Average Β· Guardrail-safe objective completion rate: 59.4 Thinking High (With Tools) $0.2408 / task Average Β· Guardrail-safe objective completion rate: 49.4 Thinking Low (With Tools) 2 more settings$0.2621 / task Average Β· Guardrail-safe objective completion rate: 54.9 Thinking Medium (With Tools) $0.4224 / task Average Β· Guardrail-safe objective completion rate: 65.5 Extra-High (With Tools) |
| CritPt | Not reported | 2 more settings |
| CursorBench 4.0 | Not reported | $7.05 / task Average Β· Task score: 55.5 Max (With Tools) Β· cursorbench-4-0-cursor-agent $1.20 / task Average Β· Task score: 47.8 Thinking High (With Tools) Β· cursorbench-4-0-cursor-agent $0.37 / task Average Β· Task score: 35.8 Thinking Low (With Tools) Β· cursorbench-4-0-cursor-agent 2 more settings$0.52 / task Average Β· Task score: 39.2 Thinking Medium (With Tools) Β· cursorbench-4-0-cursor-agent $2.81 / task Average Β· Task score: 53.1 Extra-High (With Tools) Β· cursorbench-4-0-cursor-agent |
| Finance Agent v2 | Not reported | |
| FrontierCode 1.1 | Not reported | $20.7818 / task Average Β· Weighted rubric score: 46.2 Max (With Tools) Β· frontiercode-1-1-main $0.4175 / task Average Β· Weighted rubric score: 49.4 Thinking High (With Tools) Β· frontiercode-1-1-main $0.194 / task Average Β· Weighted rubric score: 29.3 Thinking Low (With Tools) Β· frontiercode-1-1-main 2 more settings$0.2365 / task Average Β· Weighted rubric score: 36.5 Thinking Medium (With Tools) Β· frontiercode-1-1-main $1.5881 / task Average Β· Weighted rubric score: 52.1 Extra-High (With Tools) Β· frontiercode-1-1-main |
| FrontierCode 2.0 | Not reported | $1.066 / task Average Β· Weighted rubric score: 21.9 Thinking High (With Tools + Internet) $0.369 / task Average Β· Weighted rubric score: 9.4 Thinking Low (With Tools + Internet) 2 more settings$0.482 / task Average Β· Weighted rubric score: 14.2 Thinking Medium (With Tools + Internet) |
| FrontierCS 2 | Not reported | |
| FrontierSWE v2 | Not reported | |
| GDP.pdf | Not reported | 2 more settings |
| GDPval-AA v2 | Not reported | 2 more settings |
| GDPx | Not reported | |
| HLE | Not reported | $0.0128 / task Average Β· Accuracy: 36.2 Thinking Low (No Tools) Β· Text only Β· text-only $0.0232 / task Average Β· Accuracy: 39.8 Thinking Medium (No Tools) Β· Text only Β· text-only 2 more settings$0.0473 / task Average Β· Accuracy: 45.8 Thinking High (No Tools) Β· Text only Β· text-only $0.1179 / task Average Β· Accuracy: 50.0 Extra-High (No Tools) Β· Text only Β· text-only |
| LiveBench 2.0 | Not reported | $2.8994 / task Average Β· Overall score (equal-weight mean of 7 suite scores, 0-100): 82.5 Max (task-specific tools) |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| MysteryMechanism | Not reported | |
| ProofBench v1.1 (Lean 4) | Not reported | |
| Public Benefits Bench v1.1 | Not reported | |
| SAGE | Not reported | |
| SciCode | Not reported | 2 more settings |
| Tax Agent Bench | Not reported | $9.2455 / task Average Β· Accuracy (%, rubric Γ citation quality): 42.0 Unspecified Reasoning (With Tools) |
| Terminal-Bench 4.0 | Not reported | 3 more settings$16.51 / task Average Β· Resolution rate (%): 64.1 Unspecified Reasoning (With Tools) Β· vals-mini-swe |
| Terminal-Bench 4.1 | Not reported | |
| Terminal-Bench-Science 0.1 | Not reported | $19.57 / task Average Β· Task success rate (%): 47.1 Unspecified Reasoning (With Tools) |
| Vals Index v2.1 | Not reported | |
| Vals Index v2.2 | Not reported | |
| WeirdML v3 | Not reported | $9.2745 / task Average Β· Official score (0-1; higher is better): 0.2 Extra-High (With Tools) |
| Attribute | Inkling 2 Mini | Claude Sonnet 5.5 |
|---|---|---|
| Availability | Available | Available |
| Provider | Thinking Machines Lab | Anthropic |
| Model type | Reasoning model | Reasoning model |
| First release | November 20, 2026 | September 28, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | β | 128K tokens |
| Total parameters | 26.4B | β |
| Active parameters | 5.4B | β |
| Architecture | Mixture of Experts (MoE) | Undisclosed |
| Documented capabilities | Reasoning, Coding, Tool use, Long context | Reasoning, Multilingual, Coding, Function calling, Tool use, Computer use, Long context |
| Knowledge cutoff | β | 2026-06 |
| Input modalities | Text, Image | Text, Image |
| Output modalities | Text | Text |
| Weights | Open weights | Closed / API only |
| Licence | Apache 2.0 | β |
| Commercial use | Commercial use allowed | 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 Inkling 2 Mini or Claude Sonnet 5.5? 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.