Spark Desk / iFlytek Spark is a LLM from iFlytek (China).
One-line fit: best used for Speech+LLM products.
Key specifications
| Provider | iFlytek |
|---|---|
| Model | Spark Desk / iFlytek Spark |
| Type | LLM |
| Context window | Large |
| Pricing (indicative) | Product/enterprise |
| Company category | Speech + LLM |
| Region | China |
| Official site | https://www.iflytek.com |
Benchmarks and public standing
CN vertical evals
Treat scores as directional. SWE-bench, Terminal-Bench, LMSYS Arena, Artificial Analysis, and vendor cards use different harnesses and are not always comparable 1:1. Re-check the latest official model card before decisions.
What this model is good at
Speech+LLM products
This aligns with iFlytek’s broader strengths:
- Chinese speech ASR/TTS
- Education verticals
- Gov/enterprise CN
Limitations and watch-outs
Global open community
- Rate limits, regional availability, and data retention policies vary by plan.
- Agent harness quality (Cursor, Claude Code, Codex, custom tools) can change outcomes more than raw model Elo.
- Open-weight availability (if any) is separate from hosted API quality and safety filters.
Ideal users
- Product engineers shipping features that match: Speech+LLM products
- Teams standardizing on the iFlytek ecosystem
- Agent builders who need a LLM
When to pick something else
- Cheaper volume: compare lower tiers from the same lab or open Chinese/EU alternatives.
- Maximum hard SWE thoughtfulness: compare Claude Fable-class and other coding flagships.
- Giant multimodal corpora / Workspace: compare Gemini-class models.
- Self-host / open weights: Llama, Qwen, DeepSeek, GLM, Mistral open lines.
Parent company
iFlytek – China speech recognition leader; Spark Desk LLMs for education and government.
Related models from iFlytek
Data is curated for AIForumSphere model directory mid-2026 directional figures.
