Nine new models on Fikra API
Nine new inference profiles are live: Seed 2.x family, Qwen Plus, Max, 3.5 Plus and 3.6 35B. Pricing inside.
Nine new models went live on Fikra API today. They span reasoning-heavy coding workloads, high-volume batch loops, agentic front-end development, and a lower entry price for general-purpose chat.
What is new
seed-2.0-pro— strong reasoning, fits hybrid thinking modes.seed-2.0-code— agentic coding and front-end workflows.seed-2.0-lite— mid-tier, balanced price/quality.seed-2.0-mini— lowest-cost tier in the family, supports structured outputs.dola-seed-2.1-turbo— newest flagship with vision and tool use.qwen3.6-35b— balanced speed/quality at MoE routing.qwen3.5-plus— best price/quality at long context.qwen-plus— commercial workhorse.qwen-max— flagship for code and long documents.
Pricing
seed-2.0-pro: $0.625 in / $3.75 out per 1M, $0.125 cached.seed-2.0-code: $0.625 in / $3.75 out, $0.125 cached.seed-2.0-lite: $0.3125 in / $2.50 out, $0.0625 cached.seed-2.0-mini: $0.125 in / $0.50 out, $0.025 cached.dola-seed-2.1-turbo: $0.625 in / $3.125 out, $0.125 cached.qwen3.6-35b: $0.469 in / $2.81 out.qwen3.5-plus: $0.50 in / $3.00 out.qwen-plus: $0.50 in / $1.50 out.qwen-max: $2.00 in / $8.00 out.
Notes
All nine are listed in the public catalog and /v1/models right now. Cached input pricing is configured for the Seed family; Qwen tiers use standard input pricing with optional cache hits visible in the response. No changes to balance or credit are required.
James Miano
CTO & ML Engineer at Lacesse Ventures. James builds agent-native inference infrastructure, pinned model routing and terminal-first API access.
Public prices, published limits
Fikra pins model builds, publishes every rate and its limits, and lets agents create accounts, quote calls and pay from a terminal. Minimum top-up $5.