chore(pricing): Update fireworks-ai pricing

[siddharthsambharia-portkey]

🔄 Pricing Update: fireworks-ai

📊 Summary (complete_diff mode)

Change Type	Count
➕ Models added	16
🔄 Models updated (merged)	8

➕ New Models

• deepseek-v3p1
• deepseek-v3p2
• glm-4p7
• glm-5p1
• gpt-oss-120b
• gpt-oss-20b
• llama-v3p3-70b-instruct
• qwen3-8b
• qwen3p6-plus
• qwen3-vl-30b-a3b-instruct
• qwen3-vl-30b-a3b-thinking
• qwen3-embedding-8b
• flux-1-dev-fp8
• flux-1-schnell-fp8
• flux-kontext-pro
• flux-kontext-max

🔄 Updated Models

• glm-5
• kimi-k2-instruct-0905
• kimi-k2p5
• kimi-k2-thinking
• minimax-m2p1
• minimax-m2p5
• minimax-m2p7
• mixtral-8x22b-instruct

Model → Pricing Category Mapping

Named Families (exact page values)

Model ID	Display Name	Input $/1M	Output $/1M	Cache Read $/1M
deepseek-v3p1	DeepSeek V3.1	$0.56	$1.68	$0.28 (50%)
deepseek-v3p2	Deepseek V3.2	$0.56	$1.68	$0.28 (50%)
glm-4p7	GLM-4.7	$0.60	$2.20	$0.30 (50%)
glm-5	GLM-5	$1.00	$3.20	$0.20 (explicit)
glm-5p1	GLM 5.1	$1.40	$4.40	$0.26 (explicit)
gpt-oss-120b	OpenAI gpt-oss-120b	$0.15	$0.60	$0.075 (50%)
gpt-oss-20b	OpenAI gpt-oss-20b	$0.07	$0.30	$0.035 (50%)
kimi-k2-instruct-0905	Kimi K2 Instruct	$0.60	$2.50	$0.30 (50%)
kimi-k2p5	Kimi K2.5	$0.60	$3.00	$0.10 (explicit)
kimi-k2-thinking	Kimi K2 Thinking	$0.60	$2.50	$0.30 (50%)
minimax-m2p5	MiniMax-M2.5	$0.30	$1.20	$0.03 (explicit)
minimax-m2p7	MiniMax M2.7	$0.30	$1.20	$0.06 (explicit)
qwen3-vl-30b-a3b-instruct	Qwen3 VL 30B A3B	$0.15	$0.60	$0.075 (50%)
qwen3-vl-30b-a3b-thinking	Qwen3 VL 30B A3B Thinking	$0.15	$0.60	$0.075 (50%)

Tier-Based (from tier table)

Model ID	Tier	Input $/1M	Output $/1M
llama-v3p3-70b-instruct	>16B	$0.90	$0.90
minimax-m2p1	MoE 0–56B	$0.50	$0.50
mixtral-8x22b-instruct	MoE 56.1–176B	$1.20	$1.20
qwen3-8b	4B–16B	$0.20	$0.20
qwen3p6-plus	MoE 56.1–176B	$1.20	$1.20

Image Models

Model ID	Type	Price
flux-1-dev-fp8	Per-step	$0.0005/step
flux-1-schnell-fp8	Per-step	$0.00035/step
flux-kontext-pro	Per-image	$0.04/image
flux-kontext-max	Per-image	$0.08/image

Embedding Models

Model ID	Input $/1M
qwen3-embedding-8b	$0.10

Skipped

• qwen3-reranker-8b — Reranker model (excluded per skill rules)

Cache/Batch Rules Applied

• Cache read = 50% of input price (or explicit named-family value)
• Batch = 50% of serverless input AND output
• No cache_write_price set (Fireworks charges cache read only)

---

Generated by Pricing Agent on 2026-04-13