Benchmark Almanac

Spark-X2.6-4B

Spark-X2.6-4B.

iFlytek Released 22 Oct 2026 Family Spark X2.6 The lab's model card
26
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
iFlytek Oct 2026

Spark-X2.6-4B

Open weights Small

No Epoch Capabilities Index score yet
  • IFBench 75.8% #22 /70
  • SciCode 36.8% #37 /47
  • BFCL V4 71.8% #6 /24

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
IFEval 93.6% #10 of 126 Model card
IFBench 75.8% #22 of 70 Model card
Humanity's Last Exam 36.9% #64 of 147 Model card
LongBench v2 53.3% #18 of 27 Model card
GraphWalks 63.2% #16 of 21 Model card
OpenAI MRCR v2 45.8% #7 of 9 Model card
AA-LCR 1.1 67.3% #5 of 5 Model card

Math and science

BenchmarkScoreRankSource
GPQA Diamond 77.1% #129 of 284 Model card
AIME 2026 91.2% #26 of 34 Model card
HMMT 2026 81.3% #16 of 18 Model card
IMO-AnswerBench 75.2% #33 of 37 Model card
Gaokao Math 2026 134.9 #1 of 1 Model card

Coding

BenchmarkScoreRankSource
SWE-Bench Pro 54.3% #51 of 77 Model card
SWE-bench Verified 62.8% #101 of 150 Model card
SWE-bench Multilingual 64.7% #52 of 70 Model card
SciCode 36.8% #37 of 47 Model card

Agents and tools

BenchmarkScoreRankSource
BFCL V4 71.8% #6 of 24 Model card
tau2-bench 86.9% #22 of 69 Model card
Claw-Eval 69.9% #5 of 14 Model card
EnterpriseOps-Gym 33.6% #1 of 2 Model card
BrowseComp 53.1% #57 of 77 Model card
MCP Atlas 62.4% #30 of 40 Model card
tau3-bench 34.7% #11 of 12 Model card
PAgentBench 74.1% #1 of 1 Model card
SWebBench 71.7% #1 of 1 Model card
VitaBench 2.0 30.8% #1 of 1 Model card

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.