Chip performance per dollar

557 head-to-head cost-per-million-tokens comparisons across Muse Spark 1.3/Open 816B, DeepSeek V4 Pro 0813 1.6T, DeepSeek V4.1 Flash 552B, DeepSeek R1, Kimi K3.1 Pro 2.8T, Kimi K3 2.8T, Kimi K2.5/K2.6/K2.7-Code 1T, GLM 5/5.1, GLM 5.3 744B, MiniMax M3 428B, MiniMax M3.1 Flash 428B, MiniMax M2.5/M2.7, Qwen 3.8 Flash Next 176B-A6B, Qwen 3.5 397B-A17B, gpt-oss 120B, and Llama 3.3 70B. Performance is normalized by owning-hyperscaler TCO. Each page renders the cost-per-token chart and an interpolated dollars-per-million table so you can pick the cheaper SKU at any target interactivity level.

Models with AgentX data open long-context, multi-turn trace replay results. Models not yet covered by AgentX open the controlled 8K/1K workload. Each pair lists its scenario.

Muse Spark 1.3/Open 816B

28 chip pairs with cost-per-token benchmark data on Muse Spark 1.3/Open 816B

DeepSeek V4 Pro 0813 1.6T

36 chip pairs with cost-per-token benchmark data on DeepSeek V4 Pro 0813 1.6T

DeepSeek V4.1 Flash 552B

36 chip pairs with cost-per-token benchmark data on DeepSeek V4.1 Flash 552B

DeepSeek R1

55 chip pairs with cost-per-token benchmark data on DeepSeek R1

NVIDIA vs AMD

Cross-vendor cost-per-token comparisons across architecture generations

Kimi K3 2.8T

28 chip pairs with cost-per-token benchmark data on Kimi K3 2.8T

Kimi K2.5/K2.6/K2.7-Code 1T

36 chip pairs with cost-per-token benchmark data on Kimi K2.5/K2.6/K2.7-Code 1T

GLM 5.3 744B

36 chip pairs with cost-per-token benchmark data on GLM 5.3 744B

AMD vs AMD

CDNA 3 and CDNA 4 generation cost-per-token comparisons

MiniMax M3 428B

45 chip pairs with cost-per-token benchmark data on MiniMax M3 428B

MiniMax M3.1 Flash 428B

28 chip pairs with cost-per-token benchmark data on MiniMax M3.1 Flash 428B

MiniMax M2.5/M2.7

36 chip pairs with cost-per-token benchmark data on MiniMax M2.5/M2.7

Qwen 3.8 Flash Next 176B-A6B

28 chip pairs with cost-per-token benchmark data on Qwen 3.8 Flash Next 176B-A6B

Qwen 3.5 397B-A17B

55 chip pairs with cost-per-token benchmark data on Qwen 3.5 397B-A17B

gpt-oss 120B

28 chip pairs with cost-per-token benchmark data on gpt-oss 120B