Model comparison
Claude Haiku 4.5 vs Grok 4.20 Multi-Agent
Grok 4.20 Multi-Agent is the stronger model overall, scoring 46.2 to 39.5 on the Noometry Index.
Last verified . 18 shared benchmarks.
Summary
- They share 18 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 2 categories and Grok 4.20 Multi-Agent in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.20 Multi-Agent leads 43.9 to 15.1.
- The biggest single-benchmark swing is NYT Connections (extended): 14.3% for Claude Haiku 4.5 and 89.6% for Grok 4.20 Multi-Agent.
- Grok 4.20 Multi-Agent is cheaper at $1.25 / $2.50 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
- Grok 4.20 Multi-Agent accepts more context: 1M tokens versus 200K.
Side by side
| Claude Haiku 4.5 | Grok 4.20 Multi-Agent | |
|---|---|---|
| Provider | Anthropic | xAI |
| Noometry Index | 39.5 | 46.2 |
| Released | 2025-10-15 | 2026-03-09 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1M |
| Max output | 64K | 30K |
| Input $ / M tokens | $1 | $1.25 |
| Output $ / M tokens | $5 | $2.50 |
| Results tracked | 53 | 20 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude Haiku 4.5: 44.0 (#78), Grok 4.20 Multi-Agent: 43.0 (#92)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Coding | 1453 | 1457 |
| SWE-bench Verified (bash only) | 66.6% | — |
| LMArena WebDev | 1330 | — |
| SWE-bench Multilingual | 64.7% | — |
| SciCode | 43.3% | — |
| WeirdML | 45.4% | — |
| ALE-Bench | 653.48 | — |
Agentic & Tool Use Not comparable
Claude Haiku 4.5: 33.6 (#52), Grok 4.20 Multi-Agent: —
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| Terminal-Bench | 35.5% | — |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| LMArena Search | — | 1204 |
| Vending-Bench 2 | 458.89 | — |
Reasoning Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 15.1 (#320), Grok 4.20 Multi-Agent: 43.9 (#48)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| NYT Connections (extended) | 14.3% | 89.6% |
| LMArena Hard Prompts | 1420 | 1448 |
| ARC-AGI-2 | 4% | — |
| ARC-AGI-1 | 47.7% | — |
| CritPt | 0% | — |
| Chess Puzzles | 8% | — |
| DTBench | 73.6% | — |
| LMCA | 30.9% | — |
| Epoch Capabilities Index | 142.41 | — |
| ForecastBench | 61.4 | — |
Math Claude Haiku 4.5 leads
Claude Haiku 4.5: 44.9 (#78), Grok 4.20 Multi-Agent: 39.4 (#104)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Math | 1396 | 1442 |
| OTIS Mock AIME 2024-2025 | 66.7% | — |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 37.7 (#153), Grok 4.20 Multi-Agent: 40.4 (#119)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Expert | 1442 | 1445 |
| GPQA Diamond | 71.2% | — |
| SimpleQA Verified | 13.2% | — |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 26.8 (#118), Grok 4.20 Multi-Agent: 40.5 (#48)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Vision | — | 1259 |
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 49.9 (#129), Grok 4.20 Multi-Agent: 54.4 (#43)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Non-English | 1377 | 1440 |
| LMArena Chinese | 1417 | 1475 |
| LMArena French | 1408 | 1466 |
| LMArena German | 1375 | 1456 |
| LMArena Japanese | 1339 | 1405 |
| LMArena Korean | 1347 | 1416 |
| LMArena Russian | 1381 | 1457 |
| LMArena Spanish | 1420 | 1447 |
Instruction Following Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 71.4 (#149), Grok 4.20 Multi-Agent: 74.8 (#84)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Instruction Following | 1414 | 1420 |
| IFEval | 80.1% | — |
Long Context Too close to call
Claude Haiku 4.5: 43.6 (#92), Grok 4.20 Multi-Agent: 43.7 (#88)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Longer Query | 1427 | 1431 |
Writing & Preference Grok 4.20 Multi-Agent leads
Claude Haiku 4.5: 57.9 (#123), Grok 4.20 Multi-Agent: 64.0 (#59)
| Benchmark | Claude Haiku 4.5 | Grok 4.20 Multi-Agent |
|---|---|---|
| LMArena Text | 1396 | 1450 |
| LMArena Creative Writing | 1372 | 1436 |
| LMArena Multi-Turn | 1409 | 1452 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than Grok 4.20 Multi-Agent?
Grok 4.20 Multi-Agent is the stronger model overall, scoring 46.2 to 39.5 on the Noometry Index.
Which is cheaper, Claude Haiku 4.5 or Grok 4.20 Multi-Agent?
Grok 4.20 Multi-Agent is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.
Is Claude Haiku 4.5 or Grok 4.20 Multi-Agent better for coding?
They score almost the same on coding (44.0 vs 43.0); test both on your own repository before choosing.
Which has the bigger context window?
Grok 4.20 Multi-Agent does, with 1M tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and Grok 4.20 Multi-Agent share?
18 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Grok 4.20 Multi-Agent has 20.