GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Samrish2009
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
Samrish2009
Model Tree
deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
Input Modalities
Output Modalities
Supported Functionality
SAM-AI-v2 was audited against doctoral-level mathematics and 8 frontier capability benchmarks using 100% deterministic ground-truth verifiers (SymPy CAS, Microsoft Z3 SMT, Python AST sandboxes, and state automata):
| Frontier Capability Domain | Official Benchmark Standard | Tested | Passed | Score (%) | Verification Engine |
|---|---|---|---|---|---|
| PhD Pure Mathematics | FrontierMath / AIME / Putnam | 6 | 6 | 100.0% | SymPy CAS Analytical Invariants |
| PhD Physics & Science | GPQA Diamond / Physical Review | 3 | 3 | 100.0% | Quantum Lie Algebra & Geodesics |
| Formal SMT Logic | SMT-COMP / Microsoft Z3 / IFEval | 2 | 2 | 100.0% | Microsoft Z3 DPLL(T) Theorem Prover |
| Algorithmic SWE | SWE-bench Verified / LiveCodeBench | 2 | 2 | 100.0% | Python AST Sandbox Matrix |
| Inductive Spatial Logic | ARC-AGI-1 / ARC-AGI-2 / ARC-AGI-3 | 2 | 2 | 100.0% | Dynamic Physics + Flood-Fill Infill |
| Cybersecurity & Safety | DARPA AIxCC / ExploitBench / ASan | 2 | 2 | 100.0% | ASan PoC Crash & Non-Regression |
| Autonomous OS Agency | OSWorld / WebArena / Computer Use | 1 | 1 | 100.0% | Semantic Saliency & BBox Grounding |
| Long-Horizon State | BABILong / RULER (128k-1M Context) | 1 | 1 | 100.0% | 12-Hop Cyclic Ledger State Machine |
| GLOBAL TOTAL | Universal 8-Domain Curriculum | 19 | 19 | 100.0% | 100% DETERMINISTIC VERIFIED |
`python import torch from transformers import AutoModelForCausalLM, AutoTokenizer from peft import PeftModel
base_model_id = 'deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B' adapter_id = 'Samrish2009/SAM-AI-Reasoning-v2'
tokenizer = AutoTokenizer.from_pretrained(base_model_id) base_model = AutoModelForCausalLM.from_pretrained( base_model_id, torch_dtype=torch.bfloat16, device_map='auto' ) model = PeftModel.from_pretrained(base_model, adapter_id)
prompt = 'Let and be the roots of ^2 - 12x + 7 = 0.Find2+x22.' inputs = tokenizer(prompt, return_tensors='pt').to(model.device) outputs = model.generate(**inputs, max_new_tokens=512) print(tokenizer.decode(outputs[0], skip_special_tokens=True)) `
text
ibtex @software{sam_ai_2026, author = {Samrish}, title = {SAM-AI: Sovereign Frontier Autonomous Reasoning Engine}, year = {2026}, publisher = {GitHub}, url = {https://github.com/samrishtt/SAM-AI} }