GPT-4o (2024-05-13) vs Saba
GPT-4o (2024-05-13) leads Saba on the Artificial Analysis Intelligence Index, 7.3 to 6.5, and Saba costs 25.0x less per output token ($0.60 vs $15.00 per million). Updated October 23, 2026; 4 benchmarks have a result for both.
GPT-4o (2024-05-13) or Saba: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick GPT-4o (2024-05-13) for a larger context window (128K against 33K tokens) and work measured by Humanity's Last Exam (Artificial Analysis run), where it leads.
- Shared results
- 4
- GPT-4o (2024-05-13) only
- 40
- Saba only
- 0
- Like-for-like categories
- 3 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
GPT-4o (2024-05-13)
GPT-4o (2024-05-13) leads, 7.3 against 6.5.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only GPT-4o (2024-05-13) has a GPQA Diamond result, 50.2%.
Directional only
Long documents
Listed context window
GPT-4o (2024-05-13)
GPT-4o (2024-05-13) takes 128K tokens against 33K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Saba
GPT-4o (2024-05-13) $0.013, Saba $0.0005 per turn.
List prices
At a glance
- Coding index
- GPT-4o (2024-05-13) not benchmarked, Saba not benchmarked.
- Widest category
- Knowledge, 12.9 points apart on average over 1 shared benchmark.
- Widest single gap
- MMLU Pro (Artificial Analysis run): GPT-4o (2024-05-13) 74.0% against Saba 61.1%, 12.9 points apart on a 0 to 100 scale.
- Output price
- GPT-4o (2024-05-13) $15.00 against Saba $0.60 per million output tokens, 25.0x apart.
- Context window
- GPT-4o (2024-05-13) takes 128K tokens, Saba 33K.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | GPT-4o (2024-05-13) | Bars | Saba | Ties |
|---|---|---|---|---|
| Intelligence | 1 | 0 | ||
| Reasoning | 1 | 1 | ||
| Knowledge | 1 | 0 | ||
| All shared boards | 3 | 1 |
CodingDirectional onlySix of these boards feed the Vibe Coding Index4 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| AA Coding Index | 24.2 | Not measured | Only GPT-4o (2024-05-13) measured |
| LiveCodeBench (Artificial Analysis run) | 33.4% | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: coding | 1297 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: software and IT | 1297 | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 39 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| AA Intelligence Index | 7.3 | 6.5 | GPT-4o (2024-05-13) |
| Neither model has a result on the other 1 board here | |||
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| METR time horizons | 40.8% | Not measured | Only GPT-4o (2024-05-13) measured |
| BALROG | 32.3% | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 40 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| GPQA Diamond | 50.2% | Not measured | Only GPT-4o (2024-05-13) measured |
| GPQA Diamond (Artificial Analysis run) | 52.6% | 42.4% | GPT-4o (2024-05-13) |
| Humanity's Last Exam (Artificial Analysis run) | 1.8% | 4.3% | Saba |
| GPQA Diamond (Epoch AI run) | 48.9% | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: hard prompts | 1281 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: expert | 1250 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: hard prompts in English | 1293 | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 16 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index4 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| OTIS Mock AIME | 6.3% | Not measured | Only GPT-4o (2024-05-13) measured |
| MATH level 5 | 51.0% | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: math | 1284 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: mathematical work | 1284 | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 10 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| MMLU Pro (Artificial Analysis run) | 74.0% | 61.1% | GPT-4o (2024-05-13) |
| MMLU | 84.2% | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: science | 1295 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: medicine and healthcare | 1278 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: legal and government | 1314 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: business and finance | 1261 | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 14 boards here | |||
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| Neither model has a result on 9 boards here | |||
GeneralDirectional onlyPublic boards, shown as published, kept out of the index20 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| LMArena Text | 1300 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: instruction following | 1278 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: creative writing | 1291 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Vision | 1137 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: English | 1312 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: non-English | 1283 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Chinese | 1277 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: German | 1282 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Japanese | 1257 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Korean | 1234 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: French | 1304 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Spanish | 1292 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Russian | 1285 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: Polish | 1365 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: multi-turn | 1302 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: longer queries | 1288 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: writing and language | 1298 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Text: entertainment and media | 1289 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Vision: English | 1147 | Not measured | Only GPT-4o (2024-05-13) measured |
| LMArena Vision: Chinese | 1132 | Not measured | Only GPT-4o (2024-05-13) measured |
| Neither model has a result on the other 28 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | GPT-4o (2024-05-13) | Saba | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Saba costs less
Repository review
50K input + 3K output tokens
Saba costs less
Agent loop
200K input + 10K output tokens
Saba costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | OpenAI GPT-4o (2024-05-13) | Mistral Saba |
|---|---|---|
| Maker | OpenAI | Mistral AI |
| Weights | No public weights listed | No public weights listed |
| Released | May 13, 2024 | February 17, 2025 |
| Context window | 128K | 33K |
| Input $/Mtok | $5.00 | $0.20 |
| Output $/Mtok | $15.00 | $0.60 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-4o (2024-05-13) better than Saba for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick GPT-4o (2024-05-13) for a larger context window (128K against 33K tokens) and work measured by Humanity's Last Exam (Artificial Analysis run), where it leads.
Which is cheaper, GPT-4o (2024-05-13) or Saba?
GPT-4o (2024-05-13) costs $5.00 per million input tokens and $15.00 per million output tokens; Saba costs $0.20 and $0.60. Cost per solved task is known for neither of them.
Which has the bigger context window, GPT-4o (2024-05-13) or Saba?
GPT-4o (2024-05-13), with 128K tokens against 33K.
How many benchmarks is the GPT-4o (2024-05-13) vs Saba comparison based on?
4 benchmarks have a result for both models as of October 23, 2026. GPT-4o (2024-05-13)'s coding index rests on 0 independent benchmarks and Saba's on 0. Where only one model has a result, the page says so instead of naming a winner.