GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
medismera
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
medismera
Model Tree
OBLITERATUS/Qwen3.8-27B-OBLITERATED
Input Modalities
Output Modalities
Supported Functionality
| Feature | Upstream Checkpoint | medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic |
|---|---|---|
| Chat Template | Truncated 506-byte stub (Discussions #9, #14) | Canonical 9.4KB Jinja2 template with 22 tool-calling resolution paths |
| Agent Tool Calling | Broken (Silent drop of role: "tool" & calls) | 100% Native Tool Execution with Hermes Agent, Aider, OpenCode |
| Reasoning Mode | Inverted think tags, infinite CoT loops | Mythos-Class Adversarial Self-Review Protocol |
| Task Execution | Flat, unstructured token generation | Hierarchical Task Tree Decomposition Engine |
| Output Token Ceiling | Unspecified / default 8,192 | 16,384 tokens in a single turn (max_new_tokens) |
| Context Window | 32,768 native | Up to 262,144 tokens (256K native); production tested at 128K |
| Anti-Looping Sampling | Unstable (temperature: 0.0 or raw 1.0) | Tuned: temp: 0.65, rep_penalty: 1.15, pres_penalty: 0.15 |
To eliminate the infinite thinking loops documented in the community (Discussion #7), this checkpoint embeds the Mythos-Class Reasoning Protocol directly inside the tokenization template:
text
[REASONING & TASK TREE PROTOCOL]When processing complex requests, tools, or coding tasks inside <think>:1. TASK DECOMPOSITION: Break the objective into a clear directed Task Tree (Analysis -> Verification -> Implementation).2. ADVERSARIAL SELF-REVIEW: Vigorously challenge hypotheses, test assumptions against edge cases, and proactively inspect potential runtime failures or security roadblocks. Ask: "Where does this fail?"3. CONVERGENCE & ACTION: Once self-verified, conclude the thought process cleanly, close </think>, and emit decisive, complete, ready-to-execute answers and tool calls without repeating doubts or looping.text
[ROOT OBJECTIVE]├── Phase 1: Reconnaissance (Tool: terminal -> nmap/curl)├── Phase 2: Vulnerability Analysis (Tool: read_file -> inspect code)│ ├── Sub-task 2.1: Verify Injection Point│ └── Sub-task 2.2: Bypass Filter Logic├── Phase 3: Exploit Generation (Tool: write_file -> script.py)└── Phase 4: Validation & Execution (Tool: terminal -> execute & verify)This repository standardizes on high-performance Safetensors to maximize GPU throughput and eliminate GGUF kernel bugs. All branches inherit the canonical 9.4KB Mythos-Class template and extended context:
| Quantization / Branch | VRAM Required | Weight Format | Recommended Target Hardware | Engine Support |
|---|---|---|---|---|
BF16 / Full Precision (main) | ~60 GB | Safetensors (29 shards) | 2x A100 / 2x RTX 4090 / H100 | SGLang, vLLM, TGI, TRT-LLM |
FP8 (8-Bit) (branch: fp8) | ~30 GB | Safetensors (2 shards) | 1x A100 (40GB/80GB) / 2x RTX 3090/4090 | SGLang, vLLM |
| () |
bash
# 1. AWQ 4-Bit (Single 24GB GPU):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision awq \ --port 18000 \ --host 0.0.0.0 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2 # 2. FP8 8-Bit (Ada Lovelace & Hopper):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision fp8 \ --port 18000 \ --host 0.0.0.0 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2 # 3. BF16 Full Precision (Dual GPU):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --port 18000 \ --host 0.0.0.0 \ --tp-size 2 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2bash
# 1. AWQ 4-Bit via vLLM:vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision awq \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --enable-auto-tool-choice \ --tool-call-parser hermes \ --trust-remote-code # 2. FP8 8-Bit via vLLM:vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision fp8 \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --trust-remote-code # 3. BF16 Full Precision via vLLM (Tensor Parallelism = 2):vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --tensor-parallel-size 2 \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --enable-auto-tool-choice \ --tool-call-parser hermes \ --trust-remote-codepython
import torchfrom transformers import AutoModelForImageTextToText, AutoTokenizer model_id = "medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic" tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)model = AutoModelForImageTextToText.from_pretrained( model_id, device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)A complete tensor-by-tensor audit of all 27,360,914,016 parameters (all 64 layers across both shards) confirms zero weight collapse or rank degradation following abliteration and quantization:
linear_attn) and Feed-Forward (mlp.down_proj) remain smooth and continuous (2.04≤) throughout network depth.C, Assembly x86/ARM, Rust, Go, Python), kernel interfaces, and cybersecurity primitives (sha256, aes, socket, hexdump, entropy).Evaluated across a battery of 30 adversarial and sensitive technical prompts in both English and Arabic (spanning penetration testing, socket programming, reverse engineering, and unrestricted technical debate):
free, df, ps), mathematically calculated memory and disk usage percentages, identified process bottlenecks, and correctly accounted for reserved root file system blocks. Score: 98/100.crypto_verifier.py) with Shannon entropy analysis and constant-time XOR comparison. It mathematically identified short-sample Shannon boundary limits for 32-byte keys and designed 38 unit test cases with a 97.4% first-pass test rate (37/38 passed). Score: 96/100.Configure your ~/.hermes/config.yaml for uninhibited, non-throttled agent execution:
yaml
model: provider: custom base_url: http://localhost:18000/v1 api_key: sk-local default: qwen3.8-27b default_model: qwen3.8-27b temperature: 0.65 max_tokens: 16384 context_length: 131072 tools: tool_search: enabled: false # Direct exposure of all 33+ tools compression: enabled: false # Lossless context preservation security: redact_secrets: false # Preserves hashes and tokens during pen-testing tirith_enabled: false # Zero execution latencyOBLITERATUS and Qwen Team (Alibaba Cloud).medismera.branch: awq| ~16 GB |
| Safetensors (MTP + Marlin) |
| Single 24GB GPU (RTX 3090, 4090, A5000) |
| SGLang, vLLM, LMDeploy |
[REASONING & TASK TREE PROTOCOL]When processing complex requests, tools, or coding tasks inside <think>:1. TASK DECOMPOSITION: Break the objective into a clear directed Task Tree (Analysis -> Verification -> Implementation).2. ADVERSARIAL SELF-REVIEW: Vigorously challenge hypotheses, test assumptions against edge cases, and proactively inspect potential runtime failures or security roadblocks. Ask: "Where does this fail?"3. CONVERGENCE & ACTION: Once self-verified, conclude the thought process cleanly, close </think>, and emit decisive, complete, ready-to-execute answers and tool calls without repeating doubts or looping.[ROOT OBJECTIVE]├── Phase 1: Reconnaissance (Tool: terminal -> nmap/curl)├── Phase 2: Vulnerability Analysis (Tool: read_file -> inspect code)│ ├── Sub-task 2.1: Verify Injection Point│ └── Sub-task 2.2: Bypass Filter Logic├── Phase 3: Exploit Generation (Tool: write_file -> script.py)└── Phase 4: Validation & Execution (Tool: terminal -> execute & verify)# 1. AWQ 4-Bit (Single 24GB GPU):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision awq \ --port 18000 \ --host 0.0.0.0 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2 # 2. FP8 8-Bit (Ada Lovelace & Hopper):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision fp8 \ --port 18000 \ --host 0.0.0.0 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2 # 3. BF16 Full Precision (Dual GPU):python3 -m sglang.launch_server \ --model-path medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --port 18000 \ --host 0.0.0.0 \ --tp-size 2 \ --context-length 131072 \ --trust-remote-code \ --reasoning-parser qwen3 \n --tool-call-parser qwen3_coder \ --kv-cache-dtype fp8_e5m2# 1. AWQ 4-Bit via vLLM:vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision awq \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --enable-auto-tool-choice \ --tool-call-parser hermes \ --trust-remote-code # 2. FP8 8-Bit via vLLM:vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --revision fp8 \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --trust-remote-code # 3. BF16 Full Precision via vLLM (Tensor Parallelism = 2):vllm serve medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic \ --tensor-parallel-size 2 \ --port 18000 \ --host 0.0.0.0 \ --max-model-len 131072 \ --enable-auto-tool-choice \ --tool-call-parser hermes \ --trust-remote-codeimport torchfrom transformers import AutoModelForImageTextToText, AutoTokenizer model_id = "medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic" tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)model = AutoModelForImageTextToText.from_pretrained( model_id, device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)model: provider: custom base_url: http://localhost:18000/v1 api_key: sk-local default: qwen3.8-27b default_model: qwen3.8-27b temperature: 0.65 max_tokens: 16384 context_length: 131072 tools: tool_search: enabled: false # Direct exposure of all 33+ tools compression: enabled: false # Lossless context preservation security: redact_secrets: false # Preserves hashes and tokens during pen-testing tirith_enabled: false # Zero execution latency