Standard base model
Eval workspace defaultstandard
Default base model in the Attempts evaluation workspace for structured extraction and JSON tasks.
Tuning: SFT
[Models]
Browse models
Open-weight models Attempts can evaluate, tune, and deploy for your task. Inference rates are indicative for planning; training and dedicated deployment are quoted in your scoped plan. Eligibility and licensing are confirmed before training.
13 models in catalog
MiniMax M3.1 Flash
minimax-m3.1-flash
Long-context multimodal model for document-heavy workflows at low inference cost.
Tuning: SFT · DPO
GLM 5.3 Flash
glm-5.3-flash
Fast vision-language model for forms, scans, and mixed text-image inputs.
Tuning: SFT
DeepSeek V4.1 Flash
deepseek-v4.1-flash
Strong price-performance for coding, extraction, and agent-style tool loops.
Tuning: SFT · DPO
DeepSeek V4.1 Flash (1017)
deepseek-v4.1-flash-1017
Updated checkpoint for teams standardizing on the latest DeepSeek flash release.
Tuning: SFT
GLM-5.3
glm-5.3
Higher-capacity GLM tier when flash models plateau on complex reasoning tasks.
Tuning: SFT · DPO · RL
Kimi K3
kimi-k3
Frontier open model for difficult extraction, long documents, and multi-step tasks.
Tuning: SFT · DPO
Qwen3 Coder 480B A35B Instruct
qwen3-coder-480b-a35b-instruct
Large code model for repository-scale context and structured codegen deliverables.
Tuning: SFT
Qwen3.8 Flash Next FP8
qwen3.8-flash-next-fp8
Efficient Qwen flash tier for high-volume classification and short-form generation.
Tuning: SFT
OpenAI gpt-oss-20b
openai-gpt-oss-20b
Compact open-weight GPT-OSS checkpoint for cost-sensitive production workloads.
Tuning: SFT · DPO
NVIDIA Nemotron 3 Ultra NVFP4
nemotron-3-ultra-nvfp4
NVIDIA Nemotron family model for enterprise assistants and tool-use prototypes.
Tuning: SFT
MiMo-V2.6-Pro-RL
mimo-v2.6-pro-rl
RL-ready vision model when reward signals and rollout evaluators are already defined.
Tuning: SFT · RL
Ember-1
ember-1
Premium multimodal model when quality targets exceed mid-tier flash options.
Tuning: SFT