# Fireworks

HQ US · serves 18 tracked open-weights models. status page · privacy policy

models
18
verified
—
of overall top 10
5
cheapest on
1 models
vs best $
2.25×
ctx (median)
1000k
uptime
99.7%

# models

model rank $/M in $/M out $/M cache src ✓
glm-5.3 #1 $0.98 $3.08 $0.18 openrouter —
kimi-k3 #2 $2.10 $10.50 $0.21 openrouter —
glm-5.3-flash #3 $0.15 $0.50 $0.03 openrouter —
deepseek-v4-flash-0731 #6 $0.22 $0.66 $0.01 requesty —
glm-5.2 #10 $1.40 $4.40 $0.14 requesty —
minimax-m3 #12 $0.30 $1.20 $0.06 requesty —
kimi-k2.6 #15 $0.95 $4.00 $0.16 requesty —
glm-5.1 #18 $1.40 $4.40 $0.26 requesty —
kimi-k2.7-code #19 $0.95 $4.00 $0.19 requesty —
inkling #21 $1.00 $4.05 $0.17 modelsdev —
deepseek-v4-flash #22 $0.10 $0.21 $0.02 openrouter —
gpt-oss-120b #63 $0.15 $0.60 $0.01 requesty —
gpt-oss-20b #75 $0.07 $0.30 $0.04 requesty —
deepseek-v4-pro-0813 — $1.32 $3.96 $0.04 requesty —
deepseek-v4.1-flash — $0.21 $0.84 $0.00 openrouter —
deepseek-v4.1-flash-1017 — $0.21 $0.84 $0.00 openrouter —
muse-glimmer-30b — $0.35 $1.50 $0.04 requesty —
muse-spark-open — $1.25 $4.25 $0.15 openrouter —

# recent changes

when (UTC)modelchange
Oct 26 05:00 deepseek-v4-flash $/M input $0.14 → $0.10 ↓ 25.0% [source switch]
Oct 26 05:00 deepseek-v4-flash $/M output $0.28 → $0.21 ↓ 25.0% [source switch]

all providers · machine-readable: providers.json