Sarvam 105B
- Input
- $0.042/M tok
- Output
- $0.17/M tok
- Blended
- $0.074/M tok · 3:1
- Speed
- ~95tok/sest.
- Context
- 128Ktokens
Long-tail import from Artificial Analysis (2026-09-24); AA variant: Sarvam 105B (high). Price = AA median host price.
Rankings
Self-hostingEstimated
How we estimate →What it would cost to run these open weights yourself on rented GPUs, compared with the API price above.
- Hardware
- 2× H200 141GB
- FP8 weights
- Throughput
- ~7,856 tok/s
- many requests batched
- Price
- $0.11–0.88 /M tok
- busy → light use
- On your own machine
- Mac Studio M5 Max 128GB (int4, ~83 tok/s single-stream)
- single consumer GPU or Mac
Assumptions (5)
- FP8 weights (105B params, 10.3B active per token) + 25% KV-cache headroom ≈ 131 GB
- 2× H200 141GB at $3.59–$7.91/GPU-hour on-demand (2026-09-24)
- ~7,856 output tok/s aggregate at batch 282 (bandwidth-bound); MoE compute scales with active params
- Blended 3:1 input:output; prefill ~90,784 tok/s
- Low = 75% utilization at the low GPU price; high = 20% at the high price
Benchmark scores
~ italic, dashed = no published score yet, estimated from related benchmarks.
Writing & chat
IFBenchVerified
#101 of 13934.4%
- #1MiniMax-M382.9%
- #2Qwen3.8 2.4T A95B82.8%
- #99Qwen3 Coder Next35.2%
- #101Sarvam 105B34.4%
- #102INTELLECT-334.0%
artificialanalysis.ai · 2026-09-24
MMMLUEstimated
~last of 5 (estimated)below measured range (<81.3%)
No published score — estimated from related benchmarks
LMArena Text, Non-English (Elo)Estimated
~#46 of 57 (estimated)~1340 Elo
- #1Gemini 4 ArgonFew votes1512
- #2Claude Opus 5.51497
- #45Trinity Large Thinking1350
- ~#46Sarvam 105BEstimated1340
- #46gpt-oss-120b1338
No published score — estimated from related benchmarks
LMArena Text (Elo)Estimated
~#46 of 57 (estimated)~1362 Elo
- #1Gemini 4 Argon1525
- #2Claude Opus 5.51507
- #45Trinity Large Thinking1369
- ~#46Sarvam 105BEstimated1362
- #46INTELLECT-31356
No published score — estimated from related benchmarks
EQ-Bench Creative Writing v3 (Elo)Estimated
~#22 of 26 (estimated)~1434 Elo
- #1GPT-6 Astra2173
- #2Claude Fable 5.12162
- #21Gemini 3.1 Pro1491
- ~#22Sarvam 105BEstimated1434
- #22Mistral Large 31412
No published score — estimated from related benchmarks
Hard reasoning
GPQA DiamondVerified
#99 of 19273.8%
- #1GPT-6 Astra96.1%
- #2Gemini 3.8 Flash95.3%
- #99Qwen3 Next 80B A3B Instruct73.8%
- #99Sarvam 105B73.8%
- #101Qwen3 Coder Next73.7%
artificialanalysis.ai · 2026-09-24
Humanity's Last ExamVerified
#106 of 19811.0%
- #1Claude Fable 5.274.3%
- #2Claude Opus 5.561.4%
- #106gpt-oss-20b11.0%
- #106Sarvam 105B11.0%
- #109Hermes 4 - Llama-3.1 405B10.9%
artificialanalysis.ai · 2026-09-24
AIME (latest)Estimated
~#23 of 34 (estimated)~68.3%
- #1GLM-5.3 NVFP499.5%
- #2Beam 1.0 501B A23B97.7%
- #22Phi-4-reasoning-plus78.0%
- ~#23Sarvam 105BEstimated68.3%
- #23Mistral Large 338.0%
No published score — estimated from related benchmarks
AA Intelligence IndexEstimated
~#73 of 103 (estimated)~14.3
- #1Gemini 4 Carbon59.4
- #2Claude Opus 5.557.6
- #72Gemma 4 31B14.7
- ~#73Sarvam 105BEstimated14.3
- #73Mistral Medium 3.514.2
No published score — estimated from related benchmarks
Agents
τ²-bench (Telecom)Verified
#52 of 11946.8%
- #1Step 3.7 Flash98.5%
- #2DeepSeek-V4-Pro-0813-nvfp4-DSpark98.3%
- #51Solar Open 100B48.2%
- #52Sarvam 105B46.8%
- #53Nemotron 3 Nano Omni 30B A3B Reasoning45.3%
artificialanalysis.ai · 2026-09-24
OSWorld-VerifiedEstimated
~#12 of 15 (estimated)~55.7%
- #1Kimi K384.8%
- #2Qwen3.8 27B84.3%
- #11Muse Glimmer65.9%
- ~#12Sarvam 105BEstimated55.7%
- #12Nemotron 3.5 Super VL NVFP452.3%
No published score — estimated from related benchmarks
OSWorld 2.0Estimated
~#3 of 5 (estimated)~53.8%
No published score — estimated from related benchmarks
BrowseCompEstimated
~#19 of 22 (estimated)~50.1%
- #1Atria Dawn Preview92.5%
- #2GPT-6 Astra91.5%
- #18Nemotron 3.5 Bolt 74B-A9B70.0%
- ~#19Sarvam 105BEstimated50.1%
- #19Mistral Medium 3.548.6%
No published score — estimated from related benchmarks
GDPval-AA (Elo)Estimated
~#51 of 92 (estimated)~823 Elo
- #1Claude Opus 5.51866
- #2Muse Wonder Full1857
- #50Step 3.7 Flash845
- ~#51Sarvam 105BEstimated823
- #51Qwen3.5 122B A10B802
No published score — estimated from related benchmarks
AutomationBench-AAEstimated
~#47 of 85 (estimated)~7.3%
- #1Claude Fable 5.276.4%
- #2Muse Wonder Full71.3%
- #46K2 Horizon 3.7B14.7%
- ~#47Sarvam 105BEstimated7.3%
- #47Qwen3.5 397B A17B6.9%
No published score — estimated from related benchmarks
τ-Bench Banking (AA)Estimated
~#61 of 91 (estimated)~11.2%
- #1Muse Spark 1.350.5%
- #2GLM-5.350.3%
- #60K-EXAONE 2.0 080311.5%
- ~#61Sarvam 105BEstimated11.2%
- #61Devstral Small 210.7%
No published score — estimated from related benchmarks