Integration Capabilities

Available Models

Explore the ecosystem of highly capable AI models ready for integration with the GLYNNE autonomous architecture. From lightning-fast production engines to preview systems.

Open Source
Model Weights

Access our library to directly download mathematical weights from the top open-source models on the market, optimized for seamless integration onto mobile devices.

Mobile & Edge
AI Integration

High-precision, top-performance quantized models ready for native execution on iOS, Android, and embedded hardware with local inference and zero latency.

⬡GLYNNE AI ENGINEERING & AUTONOMOUS REASONING

Razonamiento de IA basado en la experiencia

Interactúa con AX, el motor de razonamiento autónomo que se ejecuta bajo la capa de control de GLYNNE, para explorar cómo funciona la autonomía empresarial segura.

New: AX 5.2 model available

Direct inference, open weights, and accelerated fine-tuning on NVIDIA H100 architecture to power secure enterprise autonomy.

/// INTEGRATION AND DEPLOYMENT STEPS
Open Source Implementation Guide
Follow these 4 fundamental steps to download, configure, integrate, and optimize our open-source models and mathematical weights across your local applications and mobile devices.
01
DISCOVERYAccess our open-source model catalog (Llama 3, GPT OSS, Mistral) and directly download quantized mathematical weights in production-ready formats (GGUF, Safetensors, or ONNX).
STEP 01 OF 04✓ READY
02
STRATEGYInitialize the runtime environment using our native SDKs and optimized libraries for Python, C++, Node.js, or mobile GPU inference engines on iOS and Android.
STEP 02 OF 04✓ READY
03
EXECUTEIntegrate models directly into your mobile devices or edge servers to execute local inference with zero latency, maintaining complete control and data privacy.
STEP 03 OF 04✓ READY
04
EVOLVEFine-tune precision hyperparameters (4-bit / 8-bit quantization), apply LoRA adapters over your custom data, and continuously scale autonomous agent performance.
STEP 04 OF 04✓ READY

💀 WHEN ALL ELSE FAILS
GETTING MODELZ RIGHT
IF IT'S ABSOLUTELY NECESSARY TO IMPLEMENT AN ON-DEVICE MODEL, MAKE SURE IT'S A GOOD ONE!
✓DO:
QUANTIZE CORRECTLY
USE 4 OR 8-BIT GGUF/AWQ QUANTIZATION TO SAVE VRAM AND BATTERY WITHOUT LOSING REASONING PRECISION.
✓DO:
LOCAL CACHING
CACHE MODEL WEIGHTS AND EMBEDDINGS LOCALLY FOR INSTANT INFERENCE AND LATENCY-FREE RUNS ON DEVICE.
✓DO:
KEEP IT SHORT
BE SHORT AND CONCISE IN YOUR PROMPTS AND KEEP CONTEXT WINDOWS FOCUSED ON SINGLE TASKS.
✓DO: ACCESSIBLE ACCESSIBLE ACCESSIBLE
THIS IS PRETTY OBVIOUS: IF YOU CAN'T MAKE IT RUN OFFLINE ON DEVICE, DON'T USE AN EDGE MODEL.
✕DON'T:
MODEL INCEPTION
AVOID NESTING RECURSIVE LLM CALLS INSIDE UNCHAINED LOOPS.
✕DON'T:
OVERPARAMETERIZE
IF YOU'RE TRYING TO RUN 70B ON A MOBILE DEVICE, YOU BETTER CONNECT TO A CLOUD SERVER.
✕DON'T:
BLOCK UI THREADS
NEVER RUN HEAVY MODEL WEIGHT INFERENCE ON MAIN UI LOOPS. THAT ONLY LEADS TO MADNESS.
✕DON'T: UNENCRYPTED WEIGHTS
NEVER STORE SENSITIVE WEIGHTS OR EMBEDDINGS WITHOUT HARDWARE-LEVEL ENCRYPTION.
✕DON'T: UNCHECKED HALLUCINATIONS
SERIOUSLY, NO ONE WANTS UNCHECKED OUTPUTS, ESPECIALLY IN PRODUCTION.
AND ALWAYS REMEMBER TO ASK, KIDS:
"WHY DOES IT HAVE TO BE A 70B MODEL WHEN AN OPTIMIZED 8B ONE WORKS?"
PIXELS BY @GLYNNE AI • DOWNLOAD MODEL WEIGHTS • COMPLETELY TRACKER FREE

/// FRONTIER API
Elite Foundation Models
The absolute best-in-class frontier models available via API on the market today. Ideal for complex reasoning, autonomous agent architectures, and high-impact tasks.
Model IDSpeed (T/s)PriceRate LimitsContextMax Comp.
GPT-4o
gpt-4o
High$5.00 in / $15.00 outTier dependent128,0004,096
Claude 3.5 Sonnet
claude-3-5-sonnet-20240620
Very High$3.00 in / $15.00 outTier dependent200,0008,192
Claude 3 Opus
claude-3-opus-20240229
Medium$15.00 in / $75.00 outTier dependent200,0004,096
Gemini 1.5 Pro
gemini-1.5-pro
High$3.50 in / $10.50 outTier dependent2,000,0008,192
Kimi (Moonshot)
moonshot-v1-128k
High$3.30 in / $9.90 outTier dependent128,0004,096
Mistral Large 2
mistral-large-latest
High$3.00 in / $9.00 outTier dependent128,0008,192
/// HIGH SPEED
Production Models (Groq Cloud)
Designed for use in demanding production environments. They meet or exceed high standards for speed, quality, and reliability.
Model IDSpeed (T/s)PriceRate LimitsContextMax Comp.
Llama 3.1 8B
llama-3.1-8b-instant
560 T/sContact SalesContact Sales131,072131,072
Llama 3.3 70B
llama-3.3-70b-versatile
280 T/sContact SalesContact Sales131,07232,768
GPT OSS 120B
openai/gpt-oss-120b
500 T/s$0.15 in / $0.60 out250K TPM / 1K RPM131,07265,536
GPT OSS 20B
openai/gpt-oss-20b
1000 T/s$0.075 in / $0.30 out250K TPM / 1K RPM131,07265,536
Whisper Large V3
whisper-large-v3
-$0.111 / hour200K ASH / 300 RPM--
Whisper Large V3 Turbo
whisper-large-v3-turbo
-$0.04 / hour400K ASH / 400 RPM--
/// MULTI-AGENT
Production Systems
A collection of models and tools that collaborate to answer complex user queries in record time.
Model IDSpeed (T/s)PriceRate LimitsContextMax Comp.
Compound
groq/compound
450 T/s-200K TPM / 200 RPM131,0728,192
Compound Mini
groq/compound-mini
450 T/s-200K TPM / 200 RPM131,0728,192
/// PREVIEW EVAL
Preview Models
Preview models are intended for evaluation purposes only and should not be used in direct production environments as they may be discontinued at short notice.
Model IDSpeed (T/s)PriceRate LimitsContextMax Comp.
Canopy Labs Orpheus Arabic
canopylabs/orpheus-arabic-saudi
-$40.00 / 1M chars50K TPM / 250 RPM4,00050,000
Canopy Labs Orpheus V1 En
canopylabs/orpheus-v1-english
-$22.00 / 1M chars50K TPM / 250 RPM4,00050,000
Prompt Guard 2 22M
meta-llama/llama-prompt-guard-2-22m
-$0.03 in / $0.03 out30K TPM / 100 RPM512512
Prompt Guard 2 86M
meta-llama/llama-prompt-guard-2-86m
-$0.04 in / $0.04 out30K TPM / 100 RPM512512
MiniMax M2.7
minimaxai/minimax-m2.7
260 T/sContact SalesContact Sales196,608131,072
Safety GPT OSS 20B
openai/gpt-oss-safeguard-20b
1000 T/s$0.075 in / $0.30 out150K TPM / 1K RPM131,07265,536
Qwen 3.6 27B
qwen/qwen3.6-27b
500 T/s$0.60 in / $3.00 out250K TPM / 1K RPM131,07216,384
Qwen 3.8 27B
qwen/qwen3.8-27b
450 T/s$0.80 in / $4.00 out250K TPM / 1K RPM131,04216,384

Model ID	Speed (T/s)	Price	Rate Limits	Context	Max Comp.
GPT-4o gpt-4o	High	$5.00 in / $15.00 out	Tier dependent	128,000	4,096
Claude 3.5 Sonnet claude-3-5-sonnet-20240620	Very High	$3.00 in / $15.00 out	Tier dependent	200,000	8,192
Claude 3 Opus claude-3-opus-20240229	Medium	$15.00 in / $75.00 out	Tier dependent	200,000	4,096
Gemini 1.5 Pro gemini-1.5-pro	High	$3.50 in / $10.50 out	Tier dependent	2,000,000	8,192
Kimi (Moonshot) moonshot-v1-128k	High	$3.30 in / $9.90 out	Tier dependent	128,000	4,096
Mistral Large 2 mistral-large-latest	High	$3.00 in / $9.00 out	Tier dependent	128,000	8,192

Model ID	Speed (T/s)	Price	Rate Limits	Context	Max Comp.
Llama 3.1 8B llama-3.1-8b-instant	560 T/s	Contact Sales	Contact Sales	131,072	131,072
Llama 3.3 70B llama-3.3-70b-versatile	280 T/s	Contact Sales	Contact Sales	131,072	32,768
GPT OSS 120B openai/gpt-oss-120b	500 T/s	$0.15 in / $0.60 out	250K TPM / 1K RPM	131,072	65,536
GPT OSS 20B openai/gpt-oss-20b	1000 T/s	$0.075 in / $0.30 out	250K TPM / 1K RPM	131,072	65,536
Whisper Large V3 whisper-large-v3	-	$0.111 / hour	200K ASH / 300 RPM	-	-
Whisper Large V3 Turbo whisper-large-v3-turbo	-	$0.04 / hour	400K ASH / 400 RPM	-	-

Model ID	Speed (T/s)	Price	Rate Limits	Context	Max Comp.
Compound groq/compound	450 T/s	-	200K TPM / 200 RPM	131,072	8,192
Compound Mini groq/compound-mini	450 T/s	-	200K TPM / 200 RPM	131,072	8,192

Model ID	Speed (T/s)	Price	Rate Limits	Context	Max Comp.
Canopy Labs Orpheus Arabic canopylabs/orpheus-arabic-saudi	-	$40.00 / 1M chars	50K TPM / 250 RPM	4,000	50,000
Canopy Labs Orpheus V1 En canopylabs/orpheus-v1-english	-	$22.00 / 1M chars	50K TPM / 250 RPM	4,000	50,000
Prompt Guard 2 22M meta-llama/llama-prompt-guard-2-22m	-	$0.03 in / $0.03 out	30K TPM / 100 RPM	512	512
Prompt Guard 2 86M meta-llama/llama-prompt-guard-2-86m	-	$0.04 in / $0.04 out	30K TPM / 100 RPM	512	512
MiniMax M2.7 minimaxai/minimax-m2.7	260 T/s	Contact Sales	Contact Sales	196,608	131,072
Safety GPT OSS 20B openai/gpt-oss-safeguard-20b	1000 T/s	$0.075 in / $0.30 out	150K TPM / 1K RPM	131,072	65,536
Qwen 3.6 27B qwen/qwen3.6-27b	500 T/s	$0.60 in / $3.00 out	250K TPM / 1K RPM	131,072	16,384
Qwen 3.8 27B qwen/qwen3.8-27b	450 T/s	$0.80 in / $4.00 out	250K TPM / 1K RPM	131,042	16,384

Not everything is Chat!

Grab a coffee. Your first step toward innovation begins by talking with AX, the AXGLYNNE artificial intelligence. Together, you will brainstorm your bottlenecks, and AX will map out how our infrastructure and team can transform those blockers into automated systems.

Available Models

Open SourceModel Weights

Mobile & EdgeAI Integration

Razonamiento de IA basado en la experiencia

New: AX 5.2 model available

DISCOVERY

STRATEGY

EXECUTE

EVOLVE

Not everything is Chat!

Open Source
Model Weights

Mobile & Edge
AI Integration