Zabio has not run its own tests on this model. Everything below is aggregated from public benchmarks, with the sources named.
Price against score in each kind of job (1) Price per 1,000 tasks, USD. Score, %. Greyed out: a cheaper model already scores as well. Ringed: the best, balanced and cheap choice.
Math, Hardest 0 25 50 75 100 $1 $10 $100
gpt-oss-20b: 40.3 +/- 6.3, $0.30
gemma-3-4b-it: 7.5 +/- 2.8, $0.30
gpt-oss-120b@deepinfra: 88.9 +/- 4.5, $0.37
gemma-3-12b-it: 16.7 +/- 4.2, $0.38
phi-4: 13.8 +/- 3.7, $0.42
deepseek-v4-flash-0731@deepinfra: 76.0 +/- 1.8, $0.45
gemma-3-27b-it: 22.5 +/- 5.3, $0.48
qwen3.5-9b@deepinfra: 42.8 +/- 6.5, $0.53
qwen3-32b: 66.9 +/- 6.0, $0.66
qwen3.5-9b@parasail: 42.8 +/- 6.5, $0.68
gemma-4-26b-a4b-it: 82.2 +/- 5.8, $0.72
qwen3-14b: 66.4 +/- 6.2, $0.72
deepseek-v4.1-flash-1017@deepinfra: 85.3 +/- 1.4, $0.74
gpt-5-nano: 21.1 +/- 2.8, $0.75
llama-4-scout-17b-16e-instruct: 3.9 +/- 2.9, $0.75
llama-3.3-70b-instruct-turbo@deepinfra: 5.1 +/- 2.4, $0.78
deepseek-v4-flash-0731: 76.0 +/- 1.8, $0.84
deepseek-v4-flash-0731@parasail: 76.0 +/- 1.8, $0.84
qwen3.5-9b: 42.8 +/- 6.5, $0.89
qwen3.5-flash-2026-02-23: 33.4 +/- 2.0, $0.90
gpt-4.1-nano: 15.0 +/- 3.0, $0.90
gpt-6-luna: 88.9 +/- 1.3, $1.05
claude-haiku-5-5: 86.2 +/- 1.4, $1.05
deepseek-flash@deepinfra: 82.9 +/- 1.4, $1.05
qwen3-30b-a3b: 62.8 +/- 6.2, $1.11
glm-5.3-flash: 74.8 +/- 1.8, $1.20
glm-5.3-flash@deepinfra: 74.8 +/- 1.8, $1.20
gemma-4-31b-it: 73.3 +/- 6.7, $1.20
gpt-oss-120b: 88.9 +/- 4.5, $1.35
gpt-4o-mini: 3.8 +/- 1.4, $1.35
llama-3.3-70b-instruct-turbo@parasail: 5.1 +/- 2.4, $1.41
qwen3.6-35b-a3b@parasail: 53.5 +/- 2.8, $1.65
qwen3.6-35b-a3b@deepinfra: 53.5 +/- 2.8, $1.73
qwen3-30b-a3b-instruct-2507: 62.2 +/- 6.5, $1.80
qwen3.5-35b-a3b@deepinfra: 54.4 +/- 6.2, $1.92
qwen3.5-35b-a3b@parasail: 54.4 +/- 6.2, $1.95
gpt-5.6-luna: 57.3 +/- 4.4, $2.40
gpt-5.4-nano: 25.7 +/- 2.9, $2.48
minimax-m3@deepinfra: 71.1 +/- 6.8, $2.49
deepseek-v4.1-flash-1017: 85.3 +/- 1.4, $2.70
deepseek-v4.1-flash-1017@parasail: 85.3 +/- 1.4, $2.70
deepseek-flash: 82.9 +/- 1.4, $2.70
deepseek-flash@alibaba-intl: 82.9 +/- 1.4, $2.70
deepseek-flash@parasail: 82.9 +/- 1.4, $2.70
deepseek-flash@together: 82.9 +/- 1.4, $2.70
minimax-m3: 71.1 +/- 6.8, $2.70
minimax-m3@parasail: 71.1 +/- 6.8, $2.70
qwen3.7-plus: 63.9 +/- 2.4, $2.88
qwen3.7-plus@alibaba-us: 63.9 +/- 2.4, $2.88
qwen3.6-flash: 37.3 +/- 2.1, $3.00
qwen3.6-flash@alibaba-us: 37.3 +/- 2.1, $3.00
gemini-3.1-flash-lite: 33.5 +/- 3.9, $3.00
qwen-plus-2025-01-25: 17.8 +/- 4.4, $3.00
qwen-plus: 1.7 +/- 0.8, $3.00
inkling-small: 68.2 +/- 2.5, $3.15
deepseek-v4-flash-0731@alibaba-intl: 76.0 +/- 1.8, $3.30
gpt-4.1-mini: 18.6 +/- 2.2, $3.60
qwen3.5-35b-a3b: 54.4 +/- 6.2, $3.75
gpt-5-mini: 34.1 +/- 2.9, $3.75
gpt-3.5-turbo: 1.1 +/- 2.7, $3.75
glm-4.7: 42.9 +/- 2.4, $3.83
deepseek-flash-fast@parasail: 82.9 +/- 1.4, $4.05
minimax-m3-fast@parasail: 71.1 +/- 6.8, $4.05
qwen3-235b-a22b-thinking-2507: 47.6 +/- 2.7, $4.14
qwen3-30b-a3b-thinking-2507: 70.3 +/- 6.0, $4.20
glm-5.2@deepinfra: 41.4 +/- 4.2, $4.39
qwen3.6-35b-a3b: 53.5 +/- 2.8, $4.50
glm-4.6: 3.8 +/- 1.1, $4.50
gemini-3.5-flash-lite: 44.6 +/- 4.2, $4.65
llama-3.3-70b-instruct-turbo: 5.1 +/- 2.4, $4.68
deepseek-r1-0528: 66.4 +/- 4.9, $4.73
mistral-large-4-0: 78.5 +/- 1.5, $5.25
glm-5.3@deepinfra: 79.9 +/- 2.2, $5.44
qwen3.6-27b@deepinfra: 62.6 +/- 2.6, $5.76
qwen3.5-397b-a17b@deepinfra: 60.1 +/- 2.7, $5.85
gemini-3-flash-preview: 59.9 +/- 1.8, $6.00
qwen3.6-plus: 52.6 +/- 1.8, $6.00
qwen3.5-397b-a17b@parasail: 60.1 +/- 2.7, $6.90
qwen3.6-27b: 62.6 +/- 2.6, $7.20
qwen3.5-397b-a17b: 60.1 +/- 2.7, $7.20
kimi-k2.6: 64.1 +/- 1.6, $7.50
deepseek-v4-pro-0813@deepinfra: 81.6 +/- 1.5, $7.80
deepseek-v4-pro@deepinfra: 46.1 +/- 4.6, $7.80
gemini-3.8-flash: 83.7 +/- 1.5, $7.88
gemini-3.7-flash: 84.4 +/- 1.6, $7.88
gemini-3.5-flash: 65.3 +/- 2.2, $7.88
gemini-3.6-flash: 71.7 +/- 3.4, $7.88
kimi-k2.7-code: 74.8 +/- 2.1, $8.85
gpt-5.4-mini: 23.6 +/- 2.7, $9.00
inkling: 61.1 +/- 2.4, $9.08
deepseek-v4-pro-0813: 81.6 +/- 1.5, $9.90
deepseek-v4-pro: 46.1 +/- 4.6, $9.90
deepseek-v4-pro@parasail: 46.1 +/- 4.6, $9.90
deepseek-v4-pro@together: 46.1 +/- 4.6, $9.90
o4-mini: 28.2 +/- 2.7, $9.90
o3-mini: 20.7 +/- 2.7, $9.90
gpt-6.1-terra: 95.4 +/- 2.6, $10.50
claude-haiku-4-5: 20.0 +/- 3.2, $10.50
glm-5.3: 79.9 +/- 2.2, $10.80
glm-5.3@alibaba-intl: 79.9 +/- 2.2, $10.80
glm-5.3@parasail: 79.9 +/- 2.2, $10.80
glm-5.1: 54.5 +/- 1.8, $10.80
glm-5.2: 41.4 +/- 4.2, $10.80
glm-5.2@alibaba-intl: 41.4 +/- 4.2, $10.80
glm-5.2@alibaba-us: 41.4 +/- 4.2, $10.80
deepseek-v4-pro-fast@parasail: 46.1 +/- 4.6, $14.85
qwen3.8-max-0902: 82.8 +/- 2.9, $15.00
qwen3.6-max-preview: 57.1 +/- 2.5, $15.60
glm-5.3-fast@parasail: 79.9 +/- 2.2, $16.20
grok-4.6: 82.6 +/- 1.4, $16.50
grok-4.7: 75.5 +/- 1.6, $16.50
o3: 29.7 +/- 2.7, $18.00
gpt-4.1: 16.6 +/- 2.2, $18.00
qwen3.7-max: 80.1 +/- 2.1, $18.75
qwen3.7-max@alibaba-us: 80.1 +/- 2.1, $18.75
gpt-5: 39.2 +/- 3.0, $18.75
gpt-5.1: 19.9 +/- 3.7, $18.75
gpt-6.1-sol: 96.8 +/- 2.6, $21.00
gemini-4-carbon: 95.3 +/- 2.7, $21.00
gpt-6-sol: 94.9 +/- 2.7, $21.00
claude-sonnet-5-5: 94.4 +/- 2.7, $21.00
gemini-4-argon: 93.0 +/- 2.7, $21.00
claude-sonnet-5: 72.8 +/- 3.3, $21.00
gpt-4o: 2.3 +/- 0.8, $22.50
gemini-3.1-pro-preview: 64.1 +/- 1.7, $24.00
gpt-5.6-terra: 66.8 +/- 4.5, $24.00
gpt-5.2: 51.8 +/- 3.1, $26.25
kimi-k3@deepinfra: 70.4 +/- 4.0, $29.93
gpt-5.4: 60.9 +/- 3.3, $30.00
kimi-k3: 70.4 +/- 4.0, $31.50
kimi-k3@moonshot: 70.4 +/- 4.0, $31.50
claude-sonnet-4-6: 51.8 +/- 3.7, $31.50
claude-sonnet-4-5: 22.8 +/- 2.8, $31.50
claude-opus-5-5: 95.6 +/- 2.6, $42.00
gpt-5.6-sol: 77.2 +/- 4.2, $42.00
claude-opus-5: 92.8 +/- 1.6, $52.50
claude-opus-4-8: 72.3 +/- 2.5, $52.50
claude-opus-4-7: 66.9 +/- 2.2, $52.50
claude-opus-4-6: 61.8 +/- 1.9, $52.50
claude-opus-4-5: 34.4 +/- 2.6, $52.50
gpt-5.5: 69.1 +/- 4.5, $60.00
gpt-4-turbo: 3.7 +/- 1.2, $75.00
claude-fable-5-2: 97.4 +/- 0.7, $84.00
gpt-6-astra: 96.8 +/- 2.6, $105
claude-fable-5-1: 95.1 +/- 2.7, $105
claude-fable-5: 92.9 +/- 1.5, $105
o1: 17.6 +/- 2.8, $135
gpt-4: 1.1 +/- 0.5, $180
gpt-oss-120b@deepinfra Cheap deepseek-v4.1-flash-1017@deepinfra Balanced qwen-plus gpt-6.1-sol Best