Tiel-Coder-35B-A3B-APEX-I-MiniPlus-V2.1-GGUF

The Definitive Frontier MoE · Efficient System RAM Offload · Full 256K Context on 24GB Workstations

[!IMPORTANT]

THE DEFINITIVE SPECIFICATION IN THE 14–15 GB CEILING

This APEX-I-MiniPlus-V2.1 release represents the specialized tensor-by-tensor configuration for sparse Mixture-of-Experts quantization within a 14–15 GB envelope. Every tensor across its 40 layers, 256 micro-experts, and integrated MTP block has been mathematically allocated to maximize reasoning precision, preserve routing behavior, and prevent avoidable CPU dequantization stalls.

[!TIP]

THE QUANTIZATION SWEET SPOT: NEAR-LOSSLESS Q6_K FIDELITY AT 3-BIT FOOTPRINT

Why APEX-I-MiniPlus V2.1 outperforms standard generic quants:

• Empirical WikiText-2 Perplexity: 7.5117 ± 0.20722 — ΔPPL +0.0517 (+0.69%) versus the approx. 7.46 BF16 baseline.
• Near-Lossless Q6_K Fidelity at Less-Than-Q3_K_M Weight: Delivers reasoning fidelity, complex instruction following, and multilingual coherence typical of Q6_K while occupying only 15.23 GB (14.18 GiB), below the 16.7 GB Q3_K_M reference.
• Zero Routing Drift: Routing matrices specified by the recipe remain in uncompressed F32, preserving the routing path across the model's MoE experts.

[!WARNING]

DO NOT CONFUSE APEX-I-MINIPLUS WITH GENERIC COMMUNITY APEX-I-MINI!

Regardless of release version, NEVER confuse handcrafted APEX-I-MiniPlus builds with generic community APEX-I-Mini releases: - Generic Community APEX-I-Mini: Uniformly compresses all core MoE experts down to aggressive 2-bit IQ2_S, leaves the sensitive token output head unarmored at 3-bit Q3_K_M, and compresses attention projections down to Q3_K. In deep reasoning models, this triggers severe perplexity spikes, syntax errors, and broken code brackets. - Handcrafted APEX-I-MiniPlus V2.1: Applies a custom tensor-by-tensor architecture that preserves specified router gates in uncompressed F32, armors the token output head in high-precision Q6_K, safeguards attention gates in Q8_0, and keeps core reasoning experts at calibrated 3-bit treatment.

[!TIP]

SYSTEM RAM INFERENCE: FULL OR PARTIAL

This APEX-I-MiniPlus release is designed for full or partial system-RAM inference. Depending on the processor, memory bandwidth, and DDR4/DDR5 configuration, generation can range from 20 to 45 tok/s. With partial GPU offload, systems that cannot fit 128K or more context entirely in VRAM can place the remaining model and context load in system RAM, maintaining stable, responsive generation at longer context lengths.

Optimization History & Transparency Notice

We maintain our previous releases publicly as a transparent engineering record of continuous optimization. Below is the exact evolutionary roadmap of our MiniPlus architectures:

[!TIP]

Deployment & System Architecture Guide

• Full GPU VRAM Offload (24GB+ VRAM, -ngl 99): Effortless full offload with native 256K context support. Blistering throughput on RTX 3090 / 4090 / 5090 GPUs.
• System RAM Streaming Specialist (DDR4/DDR5 & Massive Context): Specially engineered to run either partially or entirely out of system RAM across large or full (+160k to 256k) context windows. By replacing non-linear codebooks with linear SIMD-optimized Q3_K edge experts and upgrading shared foundation experts to Q5_K across all 40 layers, AVX2 CPU dequantization stalls are completely eliminated. Depending on your processor architecture and memory bandwidth (dual-channel DDR4 or high-speed DDR5 6000+ MT/s), streaming generation in system RAM can approach speeds remarkably close to full VRAM execution, allowing both the dedicated Q8_0 multimodal vision projector (mmproj) and the integrated Multi-Token Prediction tensors to be used by compatible runtimes for speculative decoding and the Q8_0 multimodal projector (mmproj) to be loaded in GPU VRAM for OCR while the main model weights stream effortlessly from system RAM.

Explore the complete family of APEX-I-MiniPlus models in our official collection: APEX-I-MiniPlus V2.1 Hub Collection.

Quick Navigation Index

• Model Files & Technical Specifications
• Empirical Benchmarks & Fidelity Verification
• Quality Spectrum: APEX-I-MiniPlus vs. Standard Flat Quants
• Independent Benchmark of the APEX-I-MiniPlus Family (Occamy V2 Reference)
• The 24GB Miracle: Full 256K Context Runs In VRAM!
• Hardware Throughput & Offload Benchmarks (RTX 30 / 40 / 50)
• Surgical Tensor Quantization Map (Audited from GGUF)
• Recommended Configuration & CLI Setup
• Recommended Generation Parameters (Creator Official)
• Model Inherent Behavior vs. Quantization Fidelity Notice

[!IMPORTANT]

EXPLORE THE ESTABLISHED 35B MoE MINIPLUS LINEUP

These are complementary APEX-I-MiniPlus V2.1 releases, not alternate downloads of the same model. Each receives the same tensor-by-tensor approach, integrated MTP where supported, and a design suitable for full or partial system-RAM inference. Choose the model whose native strengths best fit the work you want to do:

• Qwen3.6-35B-A3B MTP APEX-I-MiniPlus-V2.1 — a versatile frontier MoE for broad reasoning, multilingual work, agents, tool use, and multimodal tasks.
• Best for: General reasoning, agent workflows, tool calling, and flexible multimodal use.
• Qwen3.6-35B-A3B MTP APEX-I-MiniPlus-V2.1 Abliterated — the V2.1 refusal-ablated Qwen3.6 edition for users who deliberately prefer reduced refusal behavior.
• Best for: Workflows where an abliterated Qwen3.6 variant is explicitly desired.
• Ornith 1.5 APEX-I-MiniPlus-V2.1 — a software-engineering-focused MoE designed for repository-scale coding and autonomous engineering agents.
• Best for: Repository-scale development, multi-file code changes, and software-engineering agents.

These remain distinct model families and editions with their own behavior and empirical results. Pick by workload and intended alignment behavior rather than treating them as interchangeable quantization variants.

Independent Benchmark of the APEX-I-MiniPlus Family (Occamy V2 Reference)

[!NOTE] External report: zephel01 independently benchmarked Occamy V2. The benchmark below was performed on Occamy-1.0 APEX-I-MiniPlus V2, not on this specific V2.1 model. It is included as independent evidence of the broader MiniPlus quantization approach.

The APEX-I-MiniPlus quantization architecture powering this model was subjected to an extensive independent evaluation by Japanese AI researcher and evaluator zephel01 (CoolZero) on an NVIDIA RTX 5090 (32GB) workstation running llama.cpp CUDA b11027 with FlashAttention (-fa on -ctk q8_0 -ctv q8_0 -ngl 99).

The evaluation tested the APEX-I hybrid MoE engine across 348 unseeded trials on SWE-bench style multi-file Python bug-fixing tasks with hidden pytest suites (llmbench):

• L6 Multi-File Code Generation (60 tasks):
• Context 32,768 (32K): 93.3% Resolved (46/60 tasks passed 5/5 consecutive trials; 20/20 on Easy–Hard).
• Context 65,536 (65K): 90.0% Resolved (45/60 tasks passed 5/5 consecutive trials).
• Match with 25–28 GB Models: Matches or exceeds the resolution rate of full 25–28 GB models (such as Ornith-1.5 and Tiel-Coder 35B-A3B) while consuming over 10 GB less VRAM (14.6 GB vs approx. 26 GB).
• Extreme Context VRAM Scaling (The Hybrid DeltaNet SSM Advantage):
• 32K Context: 14.6 GB total VRAM allocation.
• 65K Context: 15.1 GB total VRAM allocation (only +0.5 GB VRAM added when doubling context!).
• Architectural Explanation: Because 30 of the 40 layers utilize Linear Attention / DeltaNet SSM (O(1) constant recurrence memory), only the 10 full-attention anchor layers expand the KV cache. This proves empirically that 65,536 context runs 100% in VRAM on consumer 16GB GPUs (RTX 4080 / RTX 5080) without offloading to system RAM.
• Measured Real-World Throughput: Sustained single-stream generation of approx. 247 – 251 tok/s on NVIDIA RTX 5090.

Empirical Benchmarks & Fidelity Verification

Quality Spectrum: APEX-I-MiniPlus V2.1 vs. Standard Flat Quantizations

How the handcrafted APEX-I-MiniPlus V2.1 architecture compares against standard flat quantizations in llama.cpp on 35B Mixture-of-Experts architectures:

Model Files & Technical Specifications

• Base Model: Qwen/Tiel-Coder-35B-A3B
• Parameters: 35.2B total (approx. 2.6B to 3.2B active per token)
• Architecture: 40 layers, 256 micro-experts (8 active per token) + hybrid linear attention / DeltaNet recurrent layers
• Context Length: 262,144 tokens (native 256K)

Surgical Tensor Quantization Map (Audited from GGUF)

The exact tensor breakdown below has been verified directly from the compiled binary weights:

Hardware Throughput & Offload Benchmarks (RTX 30 / 40 / 50 & RAM Streaming)

Empirically verified in Unsloth Studio & llama.cpp:

• Aggressive Hybrid Offload Profile: Hybrid offload supports reasoning-enabled generation with limited VRAM while the remaining model weights stream from system RAM.

[!NOTE]

Empirical Testbed Architecture & Desktop/Server Scaling

• Empirical Benchmark Hardware: The hybrid offload and system RAM streaming behavior documented above was measured on a consumer laptop powered by an Intel 12th Gen Alder Lake architecture featuring a hybrid design of Performance Cores (P-Cores) and Efficient Cores (E-Cores) paired with dual-channel system RAM and constrained laptop power/thermal envelopes.
• Thread Scheduling & E-Core Contention: In hybrid architectures like Alder Lake, OS thread scheduling across background E-Cores and lower single-core mobile power limits introduce memory bandwidth and thread synchronization overhead during CPU dequantization.
• Dramatic Scaling on Higher-End Processors: When running on desktop or server processors (such as modern AMD Ryzen 7000 / 9000 Zen 4/5 series or high-TDP Intel desktop platforms with dedicated performance cores, large L3 caches, and high-bandwidth dual- or quad-channel DDR5 running at 6000+ MT/s), streaming generation speeds and prefill throughput will scale dramatically higher, substantially exceeding these measured mobile numbers.

The 24GB Miracle: Full 256K Context Runs In VRAM!

Tiel-Coder-35B-A3B APEX-I-MiniPlus-V2.1 fits the entire 256K context window within 24GB VRAM:

Note: Leaves comfortable headroom for display drivers and compute buffers on standard 24GB GPUs (RTX 3090, RTX 4090, RTX 5090).

[!TIP]

Empirical 16GB GPU Verification (Single Stream / Desktop)

While theoretical multi-slot server buffers estimate approx. 16.5 GiB, independent hardware testing by zephel01 on an RTX 5090 (Occamy V2 Reference) confirmed that single-stream desktop inference consumes only 14.6 GB at 32,768 ctx and only 15.1 GB at 65,536 ctx (-ctk q8_0 -ctv q8_0 -fa on). This empirically proves that full 65K context runs completely in VRAM on 16GB cards (RTX 4080 / RTX 5080) without system RAM offload!

Recommended Configuration & Setup

llama-server.exe \
 -m Tiel-Coder-35B-A3B.APEX-I-MiniPlus-V2.1.gguf \
 --port 8080 \
 --parallel 4 \
 --flash-attn on \
 --fit on \
 -c 104960 \
 --cache-type-k q8_0 \
 --cache-type-v q8_0

Recommended Generation Parameters (Tiel Coder Official)

Sampling metadata recorded from Qwen/Tiel-Coder-35B-A3B in the completed build:

[!IMPORTANT]

Model Inherent Behavior vs. Quantization Fidelity Notice

Any behavioral nuances, stylistic tendencies, domain-specific habits, or zero-shot edge-case oversights stem entirely from the original unquantized checkpoint weights and fine-tuning distribution, NOT from the APEX-I quantization process. Handcrafted APEX-I-MiniPlus strictly preserves mathematical tensor fidelity—keeping 100% of expert routing matrices (gate_inp) in uncompressed F32 (zero router drift), armoring the token output head in Q6_K, and safeguarding attention gates in Q8_0. Empirical verification records the final GGUF perplexity at 7.5117 ± 0.20722, a ΔPPL +0.0517 (+0.69%) versus the approx. 7.46 BF16 baseline.