Model comparison
Claude Sonnet 5.5 vs Granite 4.0 Micro
Claude Sonnet 5.5 is the stronger model overall, scoring 60.5 to 28.8 on the Noometry Index. Granite 4.0 Micro costs 98× less per token, which makes it the better buy when Claude Sonnet 5.5's lead doesn't matter for your workload.
Last verified . 2 shared benchmarks.
Summary
- They share 2 benchmarks with published results for both. Claude Sonnet 5.5 scores higher in 3 categories and Granite 4.0 Micro in 0 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Sonnet 5.5 leads 88.7 to 11.2.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 100% for Claude Sonnet 5.5 and 2.8% for Granite 4.0 Micro.
- Granite 4.0 Micro is cheaper at $0.017 / $0.11 per million input/output tokens, against $2 / $10 for Claude Sonnet 5.5.
- Claude Sonnet 5.5 accepts more context: 1M tokens versus 131K.
- Granite 4.0 Micro has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 5.5 | Granite 4.0 Micro | |
|---|---|---|
| Provider | Anthropic | IBM |
| Noometry Index | 60.5 | 28.8 |
| Released | 2026-09-28 | 2025-10-02 |
| Weights | Proprietary | Open |
| Context window | 1M | 131K |
| Max output | 128K | 118K |
| Input $ / M tokens | $2 | $0.017 |
| Output $ / M tokens | $10 | $0.11 |
| Results tracked | 16 | 8 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Claude Sonnet 5.5: 65.5 (#6), Granite 4.0 Micro: —
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| FrontierCode | 52.1% | — |
| CursorBench | 55.5% | — |
| LMArena WebDev | 1699 | — |
| FrontierSWE | 61.9% | — |
| SciCode | 61% | — |
| ALE-Bench | 1,819 | — |
Agentic & Tool Use Not comparable
Claude Sonnet 5.5: 44.9 (#14), Granite 4.0 Micro: —
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| APEX-Agents | 75.5% | — |
Reasoning Claude Sonnet 5.5 leads
Claude Sonnet 5.5: 52.4 (#29), Granite 4.0 Micro: 18.5 (#261)
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| CritPt | 31.4% | — |
| Chess Puzzles | — | 0% |
| Mystery Game Puzzles | 65% | — |
Math Claude Sonnet 5.5 leads
Claude Sonnet 5.5: 88.7 (#4), Granite 4.0 Micro: 11.2 (#301)
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 100% | 2.8% |
| FrontierMath (Tiers 1-3) | 88.8% | — |
| FrontierMath Tier 4 | 80.5% | — |
| ProofBench | 100% | — |
| Omni-MATH | — | 20.9% |
Knowledge Claude Sonnet 5.5 leads
Claude Sonnet 5.5: 66.0 (#11), Granite 4.0 Micro: 9.9 (#299)
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| GPQA Diamond | 95.6% | 28.3% |
| SimpleQA Verified | 46.5% | — |
| MMLU-Pro | — | 39.5% |
| GPQA (HELM) | — | 30.7% |
Multimodal Not comparable
Claude Sonnet 5.5: 53.2 (#5), Granite 4.0 Micro: —
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| Furniture Assembly | 75% | — |
Instruction Following Not comparable
Claude Sonnet 5.5: —, Granite 4.0 Micro: 70.0 (#164)
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| IFEval | — | 84.9% |
Writing & Preference Not comparable
Claude Sonnet 5.5: —, Granite 4.0 Micro: 46.7 (#212)
| Benchmark | Claude Sonnet 5.5 | Granite 4.0 Micro |
|---|---|---|
| WildBench | — | 67% |
Frequently asked questions
Is Claude Sonnet 5.5 better than Granite 4.0 Micro?
Claude Sonnet 5.5 is the stronger model overall, scoring 60.5 to 28.8 on the Noometry Index. Granite 4.0 Micro costs 98× less per token, which makes it the better buy when Claude Sonnet 5.5's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 5.5 or Granite 4.0 Micro?
Granite 4.0 Micro is cheaper. It lists at $0.017 per million input tokens and $0.11 per million output tokens; Claude Sonnet 5.5 lists at $2 and $10.
Which has the bigger context window?
Claude Sonnet 5.5 does, with 1M tokens against 131K.
How many benchmarks do Claude Sonnet 5.5 and Granite 4.0 Micro share?
2 benchmarks have published results for both models. Claude Sonnet 5.5 has 16 scored results on Noometry and Granite 4.0 Micro has 8.